first version

sam-tool-tauri/src-tauri/.gitignore

@@ -0,0 +1,2 @@
+target
+gen/schemas

sam-tool-tauri/src-tauri/Cargo.toml

@@ -0,0 +1,35 @@
+[package]
+name = "sam-tool-tauri"
+version = "0.1.0"
+description = "SAM Tool - video frame extraction and annotation"
+edition = "2021"
+rust-version = "1.77"
+
+[lib]
+name = "sam_tool_tauri_lib"
+crate-type = ["staticlib", "cdylib", "rlib"]
+
+[build-dependencies]
+tauri-build = { version = "2", features = [] }
+
+[dependencies]
+tauri = { version = "2", features = [] }
+tauri-plugin-dialog = "2"
+tauri-plugin-fs = "2"
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+chrono = { version = "0.4", default-features = false, features = ["clock", "serde"] }
+uuid = { version = "1", features = ["v4"] }
+reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls", "multipart"] }
+base64 = "0.22"
+tokio = { version = "1", features = ["rt-multi-thread", "macros"] }
+
+[profile.dev]
+incremental = true
+
+[profile.release]
+codegen-units = 1
+lto = true
+opt-level = "s"
+panic = "abort"
+strip = true

sam-tool-tauri/src-tauri/build.rs

@@ -0,0 +1,3 @@
+fn main() {
+    tauri_build::build()
+}

sam-tool-tauri/src-tauri/capabilities/default.json

@@ -0,0 +1,11 @@
+{
+  "$schema": "../gen/schemas/desktop-schema.json",
+  "identifier": "default",
+  "description": "Default capabilities for the main window",
+  "windows": ["main"],
+  "permissions": [
+    "core:default",
+    "dialog:default",
+    "fs:default"
+  ]
+}

sam-tool-tauri/src-tauri/src/annotate.rs

@@ -0,0 +1,169 @@
+use crate::annotation::{AnnotationRecord, ImageAnnotations};
+use serde::Serialize;
+use std::path::{Path, PathBuf};
+
+#[derive(Serialize, Clone)]
+pub struct ImageEntry {
+    pub filename: String,
+    pub has_annotations: bool,
+    pub annotation_count: u32,
+    pub last_updated: Option<String>,
+}
+
+fn updated_at_of(a: &AnnotationRecord) -> &str {
+    match a {
+        AnnotationRecord::Bbox { updated_at, .. } => updated_at,
+        AnnotationRecord::Polygon { updated_at, .. } => updated_at,
+    }
+}
+
+fn is_image(name: &str) -> bool {
+    let lower = name.to_lowercase();
+    lower.ends_with(".jpg") || lower.ends_with(".jpeg") || lower.ends_with(".png")
+}
+
+fn sibling_json(image_path: &Path) -> PathBuf {
+    image_path.with_extension("json")
+}
+
+pub fn list_images(folder: &Path) -> Result<Vec<ImageEntry>, String> {
+    if !folder.exists() {
+        return Err(format!("folder not found: {}", folder.display()));
+    }
+    if !folder.is_dir() {
+        return Err(format!("not a directory: {}", folder.display()));
+    }
+    let mut out = Vec::new();
+    let entries = std::fs::read_dir(folder)
+        .map_err(|e| format!("read_dir {}: {e}", folder.display()))?;
+    for entry in entries {
+        let entry = entry.map_err(|e| format!("dirent: {e}"))?;
+        let path = entry.path();
+        if !path.is_file() {
+            continue;
+        }
+        let Some(name) = path.file_name().and_then(|n| n.to_str()) else {
+            continue;
+        };
+        if !is_image(name) {
+            continue;
+        }
+        let json_path = sibling_json(&path);
+        let (annotation_count, last_updated) = if json_path.exists() {
+            match std::fs::read_to_string(&json_path)
+                .ok()
+                .and_then(|t| serde_json::from_str::<ImageAnnotations>(&t).ok())
+            {
+                Some(ia) => {
+                    let latest = ia
+                        .annotations
+                        .iter()
+                        .map(|a| updated_at_of(a).to_string())
+                        .max();
+                    (ia.annotations.len() as u32, latest)
+                }
+                None => (0, None),
+            }
+        } else {
+            (0, None)
+        };
+        out.push(ImageEntry {
+            filename: name.to_string(),
+            has_annotations: annotation_count > 0,
+            annotation_count,
+            last_updated,
+        });
+    }
+    out.sort_by(|a, b| a.filename.cmp(&b.filename));
+    Ok(out)
+}
+
+pub fn read_image_bytes(path: &Path) -> Result<Vec<u8>, String> {
+    if !path.exists() {
+        return Err(format!("file not found: {}", path.display()));
+    }
+    std::fs::read(path).map_err(|e| format!("read {}: {e}", path.display()))
+}
+
+pub fn load_image_annotations(
+    image_path: &Path,
+) -> Result<Option<ImageAnnotations>, String> {
+    let json_path = sibling_json(image_path);
+    if !json_path.exists() {
+        return Ok(None);
+    }
+    let text = std::fs::read_to_string(&json_path)
+        .map_err(|e| format!("read {}: {e}", json_path.display()))?;
+    let ia: ImageAnnotations =
+        serde_json::from_str(&text).map_err(|e| format!("parse sibling JSON: {e}"))?;
+    Ok(Some(ia))
+}
+
+pub fn load_classes(folder: &Path) -> Result<Vec<String>, String> {
+    let path = folder.join("classes.txt");
+    if !path.exists() {
+        return Ok(Vec::new());
+    }
+    let text = std::fs::read_to_string(&path)
+        .map_err(|e| format!("read {}: {e}", path.display()))?;
+    Ok(text
+        .lines()
+        .map(|l| l.trim().to_string())
+        .filter(|l| !l.is_empty())
+        .collect())
+}
+
+/// Delete a batch of image files along with their sibling annotation JSONs.
+/// Returns the count of images actually removed. Missing images are skipped;
+/// any hard I/O error aborts with what succeeded so far reflected in the log.
+pub fn delete_images(paths: &[String]) -> Result<u32, String> {
+    let mut removed: u32 = 0;
+    for p in paths {
+        let path = Path::new(p);
+        if !path.is_file() {
+            continue;
+        }
+        // Only allow image files — guard against a stray path that would
+        // wipe a JSON or classes.txt by name collision.
+        let name = path
+            .file_name()
+            .and_then(|n| n.to_str())
+            .unwrap_or_default();
+        if !is_image(name) {
+            return Err(format!("refusing to delete non-image: {}", path.display()));
+        }
+        std::fs::remove_file(path)
+            .map_err(|e| format!("remove {}: {e}", path.display()))?;
+        let json_path = sibling_json(path);
+        if json_path.exists() {
+            let _ = std::fs::remove_file(&json_path);
+        }
+        removed += 1;
+    }
+    Ok(removed)
+}
+
+pub fn append_class(folder: &Path, name: &str) -> Result<Vec<String>, String> {
+    let trimmed = name.trim();
+    if trimmed.is_empty() {
+        return Err("class name cannot be empty".into());
+    }
+    if trimmed.len() > 96 {
+        return Err("class name too long".into());
+    }
+    std::fs::create_dir_all(folder)
+        .map_err(|e| format!("mkdir {}: {e}", folder.display()))?;
+    let mut classes = load_classes(folder)?;
+    if classes.iter().any(|n| n == trimmed) {
+        return Err(format!("class '{trimmed}' already exists"));
+    }
+    classes.push(trimmed.to_string());
+    let mut text = String::new();
+    for c in &classes {
+        text.push_str(c);
+        text.push('\n');
+    }
+    let path = folder.join("classes.txt");
+    std::fs::write(&path, text).map_err(|e| format!("write {}: {e}", path.display()))?;
+    Ok(classes)
+}

sam-tool-tauri/src-tauri/src/annotation.rs

@@ -0,0 +1,43 @@
+use serde::{Deserialize, Serialize};
+
+#[derive(Serialize, Deserialize, Clone)]
+pub struct ImageAnnotations {
+    pub version: u32,
+    pub image: String,
+    pub width: u32,
+    pub height: u32,
+    /// Absolute path to the source video the frame was extracted from.
+    /// Optional so older JSONs without this field still round-trip cleanly.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub source_video: Option<String>,
+    pub annotations: Vec<AnnotationRecord>,
+}
+
+/// Internally-tagged enum. Both variants share id/class/author/timestamps and
+/// differ only in the shape data (bbox | points). Using a single enum with
+/// `#[serde(tag = "type")]` gives clean round-trip serde (flatten + tagged
+/// enums together don't deserialize reliably).
+#[derive(Serialize, Deserialize, Clone)]
+#[serde(tag = "type")]
+pub enum AnnotationRecord {
+    #[serde(rename = "bbox")]
+    Bbox {
+        id: String,
+        class_id: i64,
+        class_name: String,
+        bbox: [f64; 4],
+        author: String,
+        created_at: String,
+        updated_at: String,
+    },
+    #[serde(rename = "polygon")]
+    Polygon {
+        id: String,
+        class_id: i64,
+        class_name: String,
+        points: Vec<[f64; 2]>,
+        author: String,
+        created_at: String,
+        updated_at: String,
+    },
+}

sam-tool-tauri/src-tauri/src/coco.rs

@@ -0,0 +1,359 @@
+use serde::{Deserialize, Serialize};
+use std::collections::{BTreeSet, HashMap};
+use std::path::Path;
+
+#[derive(Deserialize)]
+struct CocoFile {
+    images: Vec<CocoImage>,
+    categories: Vec<CocoCategory>,
+    annotations: Vec<CocoAnnotation>,
+}
+
+#[derive(Deserialize)]
+struct CocoImage {
+    id: i64,
+    #[serde(default)]
+    file_name: String,
+}
+
+#[derive(Deserialize)]
+struct CocoCategory {
+    id: i64,
+    name: String,
+}
+
+#[derive(Deserialize)]
+struct CocoAnnotation {
+    image_id: i64,
+    category_id: i64,
+    #[serde(default)]
+    bbox: Option<Vec<f64>>,
+    #[serde(default)]
+    segmentation: Option<serde_json::Value>,
+}
+
+#[derive(Serialize, Deserialize, Clone)]
+#[serde(tag = "type")]
+pub enum Shape {
+    #[serde(rename = "bbox")]
+    Bbox {
+        class_id: i64,
+        class_name: String,
+        bbox: [f64; 4],
+    },
+    #[serde(rename = "polygon")]
+    Polygon {
+        class_id: i64,
+        class_name: String,
+        points: Vec<[f64; 2]>,
+    },
+}
+
+#[derive(Serialize, Deserialize, Clone)]
+pub struct Category {
+    pub id: i64,
+    pub name: String,
+}
+
+#[derive(Serialize, Clone, Default)]
+pub struct ImportStats {
+    pub total_annotations: usize,
+    pub frame_count: usize,
+    pub categories: usize,
+    pub skipped_rle: usize,
+    pub skipped_no_image: usize,
+    /// Filled in by the import — "coco" or "sam".
+    pub source_format: String,
+}
+
+#[derive(Serialize, Clone)]
+pub struct ImportedCoco {
+    pub categories: Vec<Category>,
+    /// Map of frame_idx (as string, per JSON) -> shapes on that frame.
+    pub frames: HashMap<String, Vec<Shape>>,
+    pub stats: ImportStats,
+}
+
+enum Format {
+    LabelStudioCoco,
+    SamAnno,
+    Unknown,
+}
+
+fn detect_format(v: &serde_json::Value) -> Format {
+    let Some(obj) = v.as_object() else {
+        return Format::Unknown;
+    };
+    if obj.contains_key("images") && obj.contains_key("annotations") {
+        return Format::LabelStudioCoco;
+    }
+    if !obj.is_empty()
+        && obj
+            .keys()
+            .all(|k| !k.is_empty() && k.chars().all(|c| c.is_ascii_digit()))
+        && obj.values().all(|v| v.is_object())
+    {
+        return Format::SamAnno;
+    }
+    Format::Unknown
+}
+
+pub fn import(path: &Path) -> Result<ImportedCoco, String> {
+    let text = std::fs::read_to_string(path)
+        .map_err(|e| format!("read {}: {e}", path.display()))?;
+    let root: serde_json::Value =
+        serde_json::from_str(&text).map_err(|e| format!("parse JSON: {e}"))?;
+
+    match detect_format(&root) {
+        Format::LabelStudioCoco => {
+            let coco: CocoFile = serde_json::from_value(root)
+                .map_err(|e| format!("parse COCO: {e}"))?;
+            Ok(build_from_coco(coco))
+        }
+        Format::SamAnno => build_from_sam_anno(&root),
+        Format::Unknown => Err(
+            "unrecognized annotation format — expected Label Studio COCO or SAM anno JSON"
+                .into(),
+        ),
+    }
+}
+
+// ---- Label Studio COCO -----------------------------------------------------
+
+fn build_from_coco(coco: CocoFile) -> ImportedCoco {
+    let mut image_frame: HashMap<i64, u64> = HashMap::with_capacity(coco.images.len());
+    for img in &coco.images {
+        let frame_idx =
+            trailing_int(&img.file_name).unwrap_or_else(|| img.id.max(0) as u64);
+        image_frame.insert(img.id, frame_idx);
+    }
+
+    let cat_name: HashMap<i64, String> = coco
+        .categories
+        .iter()
+        .map(|c| (c.id, c.name.clone()))
+        .collect();
+
+    let mut frames: HashMap<String, Vec<Shape>> = HashMap::new();
+    let mut stats = ImportStats {
+        categories: coco.categories.len(),
+        source_format: "coco".into(),
+        ..Default::default()
+    };
+
+    for ann in &coco.annotations {
+        stats.total_annotations += 1;
+        let Some(&frame_idx) = image_frame.get(&ann.image_id) else {
+            stats.skipped_no_image += 1;
+            continue;
+        };
+        let class_name = cat_name
+            .get(&ann.category_id)
+            .cloned()
+            .unwrap_or_else(|| format!("class_{}", ann.category_id));
+        let key = frame_idx.to_string();
+        let bucket = frames.entry(key).or_default();
+
+        if let Some(seg) = &ann.segmentation {
+            match seg {
+                serde_json::Value::Array(rings) => {
+                    for ring in rings {
+                        if let Some(flat) = ring.as_array() {
+                            let points = parse_flat_polygon(flat);
+                            if points.len() >= 3 {
+                                bucket.push(Shape::Polygon {
+                                    class_id: ann.category_id,
+                                    class_name: class_name.clone(),
+                                    points,
+                                });
+                            }
+                        }
+                    }
+                }
+                serde_json::Value::Object(_) => {
+                    stats.skipped_rle += 1;
+                }
+                _ => {}
+            }
+        }
+
+        if let Some(bbox) = &ann.bbox {
+            if bbox.len() == 4 {
+                bucket.push(Shape::Bbox {
+                    class_id: ann.category_id,
+                    class_name,
+                    bbox: [bbox[0], bbox[1], bbox[2], bbox[3]],
+                });
+            }
+        }
+    }
+
+    stats.frame_count = frames.len();
+    let categories = coco
+        .categories
+        .into_iter()
+        .map(|c| Category {
+            id: c.id,
+            name: c.name,
+        })
+        .collect();
+
+    ImportedCoco {
+        categories,
+        frames,
+        stats,
+    }
+}
+
+// ---- Custom SAM annotation format ------------------------------------------
+//
+// { "<frame_idx>": { "<class_name>": [[id, [tl_x,tl_y], [br_x,br_y],
+//                                       [[[x,y], ...], ...rings]]] } }
+// Bbox is redundant with the polygon's tight bounding box, so we emit only
+// polygons. Class IDs are assigned 0-indexed by sorted class name.
+fn build_from_sam_anno(root: &serde_json::Value) -> Result<ImportedCoco, String> {
+    let obj = root
+        .as_object()
+        .ok_or_else(|| "SAM anno root is not an object".to_string())?;
+
+    let mut class_names = BTreeSet::<String>::new();
+    let mut frames: HashMap<String, Vec<Shape>> = HashMap::new();
+    let mut stats = ImportStats {
+        source_format: "sam".into(),
+        ..Default::default()
+    };
+
+    for (frame_key, frame_val) in obj {
+        let Some(cls_obj) = frame_val.as_object() else {
+            continue;
+        };
+        let mut shapes: Vec<Shape> = Vec::new();
+        for (class_name, items) in cls_obj {
+            class_names.insert(class_name.clone());
+            let Some(items) = items.as_array() else {
+                continue;
+            };
+            for item in items {
+                let Some(arr) = item.as_array() else {
+                    continue;
+                };
+                if arr.len() < 4 {
+                    continue;
+                }
+                stats.total_annotations += 1;
+
+                // arr[3] is a list of polygon rings.
+                let Some(rings) = arr[3].as_array() else {
+                    continue;
+                };
+                for ring in rings {
+                    let Some(pts) = ring.as_array() else {
+                        continue;
+                    };
+                    let mut points = Vec::with_capacity(pts.len());
+                    for p in pts {
+                        let Some(pa) = p.as_array() else {
+                            continue;
+                        };
+                        let (Some(x), Some(y)) = (
+                            pa.first().and_then(|v| v.as_f64()),
+                            pa.get(1).and_then(|v| v.as_f64()),
+                        ) else {
+                            continue;
+                        };
+                        points.push([x, y]);
+                    }
+                    if points.len() >= 3 {
+                        shapes.push(Shape::Polygon {
+                            class_id: 0,
+                            class_name: class_name.clone(),
+                            points,
+                        });
+                    }
+                }
+            }
+        }
+        if !shapes.is_empty() {
+            frames.insert(frame_key.clone(), shapes);
+        }
+    }
+
+    // Assign class_ids from sorted class names.
+    let categories: Vec<Category> = class_names
+        .iter()
+        .enumerate()
+        .map(|(i, name)| Category {
+            id: i as i64,
+            name: name.clone(),
+        })
+        .collect();
+    let name_to_id: HashMap<String, i64> = categories
+        .iter()
+        .map(|c| (c.name.clone(), c.id))
+        .collect();
+
+    for shapes in frames.values_mut() {
+        for shape in shapes.iter_mut() {
+            let (cid, cname) = match shape {
+                Shape::Bbox {
+                    class_id,
+                    class_name,
+                    ..
+                }
+                | Shape::Polygon {
+                    class_id,
+                    class_name,
+                    ..
+                } => (class_id, class_name),
+            };
+            if let Some(&id) = name_to_id.get(cname) {
+                *cid = id;
+            }
+        }
+    }
+
+    stats.categories = categories.len();
+    stats.frame_count = frames.len();
+
+    Ok(ImportedCoco {
+        categories,
+        frames,
+        stats,
+    })
+}
+
+// ---- helpers ---------------------------------------------------------------
+
+fn trailing_int(file_name: &str) -> Option<u64> {
+    let stem = Path::new(file_name)
+        .file_stem()
+        .and_then(|s| s.to_str())
+        .unwrap_or(file_name);
+    let digits: String = stem
+        .chars()
+        .rev()
+        .take_while(|c| c.is_ascii_digit())
+        .collect::<Vec<_>>()
+        .into_iter()
+        .rev()
+        .collect();
+    if digits.is_empty() {
+        None
+    } else {
+        digits.parse().ok()
+    }
+}
+
+fn parse_flat_polygon(flat: &[serde_json::Value]) -> Vec<[f64; 2]> {
+    let mut out = Vec::with_capacity(flat.len() / 2);
+    let mut i = 0;
+    while i + 1 < flat.len() {
+        let x = flat[i].as_f64();
+        let y = flat[i + 1].as_f64();
+        if let (Some(x), Some(y)) = (x, y) {
+            out.push([x, y]);
+        }
+        i += 2;
+    }
+    out
+}

sam-tool-tauri/src-tauri/src/decoder.rs

@@ -0,0 +1,351 @@
+use crate::video::VideoInfo;
+use std::collections::{HashMap, VecDeque};
+use std::io::{BufRead, BufReader};
+use std::path::PathBuf;
+use std::process::{Child, ChildStdout, Command, Stdio};
+use std::sync::{Arc, Condvar, Mutex};
+use std::thread::{self, JoinHandle};
+use std::time::Duration;
+
+pub const DEFAULT_CACHE_CAPACITY: usize = 500;
+pub const MIN_CACHE_CAPACITY: usize = 200;
+pub const MAX_CACHE_CAPACITY: usize = 2000;
+const READ_BUF_BYTES: usize = 256 * 1024;
+/// If the requested frame is this far ahead of the worker, kill + reseek.
+const SEEK_AHEAD_THRESHOLD: u64 = 60;
+
+struct FrameCache {
+    map: HashMap<u64, Vec<u8>>,
+    order: VecDeque<u64>,
+    capacity: usize,
+}
+
+impl FrameCache {
+    fn new(capacity: usize) -> Self {
+        Self {
+            map: HashMap::new(),
+            order: VecDeque::new(),
+            capacity,
+        }
+    }
+
+    fn contains(&self, idx: u64) -> bool {
+        self.map.contains_key(&idx)
+    }
+
+    fn get(&mut self, idx: u64) -> Option<Vec<u8>> {
+        let bytes = self.map.get(&idx)?.clone();
+        if let Some(pos) = self.order.iter().position(|&i| i == idx) {
+            self.order.remove(pos);
+        }
+        self.order.push_back(idx);
+        Some(bytes)
+    }
+
+    fn insert(&mut self, idx: u64, bytes: Vec<u8>) {
+        if self.map.contains_key(&idx) {
+            if let Some(pos) = self.order.iter().position(|&i| i == idx) {
+                self.order.remove(pos);
+            }
+        } else if self.map.len() >= self.capacity {
+            if let Some(oldest) = self.order.pop_front() {
+                self.map.remove(&oldest);
+            }
+        }
+        self.map.insert(idx, bytes);
+        self.order.push_back(idx);
+    }
+
+    fn set_capacity(&mut self, new_cap: usize) {
+        self.capacity = new_cap.clamp(MIN_CACHE_CAPACITY, MAX_CACHE_CAPACITY);
+        while self.map.len() > self.capacity {
+            if let Some(oldest) = self.order.pop_front() {
+                self.map.remove(&oldest);
+            } else {
+                break;
+            }
+        }
+    }
+}
+
+struct SharedState {
+    /// Owned by shared state (not the worker closure) so stop_worker can
+    /// kill ffmpeg directly and unblock a pipe read that's waiting for the
+    /// first post-seek frame.
+    child: Option<Child>,
+    cache: FrameCache,
+    /// Frame the worker's pipe will emit next.
+    next_idx: u64,
+    /// Most recent frame the frontend requested — used to bound lookahead.
+    reader_idx: u64,
+    /// Worker reads this many frames past reader_idx before parking.
+    lookahead: u64,
+    eof: bool,
+    error: Option<String>,
+    shutdown: bool,
+}
+
+struct Shared {
+    state: Mutex<SharedState>,
+    cond: Condvar,
+}
+
+pub struct DecoderSession {
+    path: PathBuf,
+    info: VideoInfo,
+    shared: Arc<Shared>,
+    worker: Option<JoinHandle<()>>,
+}
+
+impl DecoderSession {
+    pub fn open(path: PathBuf, info: VideoInfo) -> Self {
+        Self {
+            path,
+            info,
+            shared: Arc::new(Shared {
+                state: Mutex::new(SharedState {
+                    child: None,
+                    cache: FrameCache::new(DEFAULT_CACHE_CAPACITY),
+                    next_idx: 0,
+                    reader_idx: 0,
+                    lookahead: (DEFAULT_CACHE_CAPACITY as u64) / 2,
+                    eof: false,
+                    error: None,
+                    shutdown: false,
+                }),
+                cond: Condvar::new(),
+            }),
+            worker: None,
+        }
+    }
+
+    pub fn info(&self) -> &VideoInfo {
+        &self.info
+    }
+
+    pub fn set_cache_capacity(&self, capacity: usize) {
+        let clamped = capacity.clamp(MIN_CACHE_CAPACITY, MAX_CACHE_CAPACITY);
+        let mut st = self.shared.state.lock().unwrap();
+        st.cache.set_capacity(clamped);
+        st.lookahead = (clamped as u64) / 2;
+    }
+
+    pub fn decode_at(&mut self, idx: u64) -> Result<Vec<u8>, String> {
+        if idx >= self.info.total_frames {
+            return Err(format!(
+                "frame {} out of range (total {})",
+                idx, self.info.total_frames
+            ));
+        }
+
+        // Update reader cursor, check cache, note whether a seek is needed.
+        let need_seek = {
+            let mut st = self.shared.state.lock().unwrap();
+            st.reader_idx = idx;
+            if let Some(bytes) = st.cache.get(idx) {
+                self.shared.cond.notify_all();
+                return Ok(bytes);
+            }
+            self.worker.is_none()
+                || idx < st.next_idx
+                || idx > st.next_idx.saturating_add(SEEK_AHEAD_THRESHOLD)
+        };
+
+        if need_seek {
+            self.seek_and_start_worker(idx)?;
+        } else {
+            self.shared.cond.notify_all();
+        }
+
+        // Wait for the worker to produce idx.
+        let mut st = self.shared.state.lock().unwrap();
+        loop {
+            if st.cache.contains(idx) {
+                return st
+                    .cache
+                    .get(idx)
+                    .ok_or_else(|| "frame vanished from cache".to_string());
+            }
+            if let Some(e) = &st.error {
+                return Err(e.clone());
+            }
+            if st.eof {
+                return Err(format!("frame {} not produced (EOF)", idx));
+            }
+            if st.shutdown {
+                return Err("decoder shut down".into());
+            }
+            st = self.shared.cond.wait(st).unwrap();
+        }
+    }
+
+    fn seek_and_start_worker(&mut self, idx: u64) -> Result<(), String> {
+        self.stop_worker();
+
+        let (child, stdout) = spawn_ffmpeg(&self.path, &self.info, idx)?;
+
+        {
+            let mut st = self.shared.state.lock().unwrap();
+            st.child = Some(child);
+            st.next_idx = idx;
+            st.reader_idx = idx;
+            st.eof = false;
+            st.error = None;
+            st.shutdown = false;
+        }
+        self.shared.cond.notify_all();
+
+        let shared = self.shared.clone();
+        let handle = thread::spawn(move || {
+            worker_loop(stdout, shared);
+        });
+        self.worker = Some(handle);
+        Ok(())
+    }
+
+    fn stop_worker(&mut self) {
+        // Take ownership of the child out of shared state and kill it FIRST so
+        // a pipe read in the worker unblocks with EOF. Then join the worker.
+        let child_opt = {
+            let mut st = self.shared.state.lock().unwrap();
+            st.shutdown = true;
+            st.child.take()
+        };
+        if let Some(mut child) = child_opt {
+            let _ = child.kill();
+            let _ = child.wait();
+        }
+        self.shared.cond.notify_all();
+
+        if let Some(handle) = self.worker.take() {
+            let _ = handle.join();
+        }
+
+        let mut st = self.shared.state.lock().unwrap();
+        st.shutdown = false;
+        st.eof = false;
+        st.error = None;
+    }
+}
+
+impl Drop for DecoderSession {
+    fn drop(&mut self) {
+        self.stop_worker();
+    }
+}
+
+fn spawn_ffmpeg(
+    path: &PathBuf,
+    info: &VideoInfo,
+    idx: u64,
+) -> Result<(Child, BufReader<ChildStdout>), String> {
+    let path_str = path
+        .to_str()
+        .ok_or_else(|| "non-utf8 path".to_string())?;
+    let t = idx as f64 / info.fps;
+    let mut cmd = Command::new("ffmpeg");
+    cmd.args(["-v", "error", "-threads", "0"]);
+    if t > 1.0 {
+        cmd.args(["-ss", &format!("{:.3}", t - 1.0)]);
+        cmd.args(["-i", path_str]);
+        cmd.args(["-ss", "1.0"]);
+    } else {
+        cmd.args(["-i", path_str]);
+        cmd.args(["-ss", &format!("{:.3}", t)]);
+    }
+    cmd.args([
+        "-an",
+        "-q:v",
+        "3",
+        "-f",
+        "image2pipe",
+        "-vcodec",
+        "mjpeg",
+        "pipe:1",
+    ]);
+    cmd.stdout(Stdio::piped()).stderr(Stdio::piped());
+    let mut child = cmd.spawn().map_err(|e| format!("ffmpeg spawn: {e}"))?;
+    let stdout = child
+        .stdout
+        .take()
+        .ok_or_else(|| "ffmpeg stdout unavailable".to_string())?;
+    Ok((child, BufReader::with_capacity(READ_BUF_BYTES, stdout)))
+}
+
+fn worker_loop(mut stdout: BufReader<ChildStdout>, shared: Arc<Shared>) {
+    loop {
+        // Backpressure: wait if we're already far enough ahead.
+        {
+            let st = shared.state.lock().unwrap();
+            if st.shutdown {
+                return;
+            }
+            let ahead = st.next_idx.saturating_sub(st.reader_idx);
+            if ahead >= st.lookahead {
+                let (st2, _) = shared
+                    .cond
+                    .wait_timeout(st, Duration::from_millis(100))
+                    .unwrap();
+                if st2.shutdown {
+                    return;
+                }
+                continue;
+            }
+        }
+
+        match read_jpeg_bytes(&mut stdout) {
+            Ok(bytes) => {
+                let mut st = shared.state.lock().unwrap();
+                if st.shutdown {
+                    return;
+                }
+                let idx = st.next_idx;
+                st.cache.insert(idx, bytes);
+                st.next_idx = idx + 1;
+                shared.cond.notify_all();
+            }
+            Err(e) => {
+                let mut st = shared.state.lock().unwrap();
+                if e.contains("pipe closed") || e.contains("EOF") {
+                    st.eof = true;
+                } else {
+                    st.error = Some(e);
+                }
+                shared.cond.notify_all();
+                return;
+            }
+        }
+    }
+}
+
+fn read_jpeg_bytes(reader: &mut BufReader<ChildStdout>) -> Result<Vec<u8>, String> {
+    let mut out = Vec::with_capacity(64 * 1024);
+    let mut last = 0u8;
+    let mut in_jpeg = false;
+    loop {
+        let buf = reader.fill_buf().map_err(|e| format!("pipe read: {e}"))?;
+        if buf.is_empty() {
+            return Err("ffmpeg pipe closed before frame completed".into());
+        }
+        let mut consumed = 0usize;
+        for &b in buf {
+            consumed += 1;
+            if !in_jpeg {
+                if last == 0xFF && b == 0xD8 {
+                    in_jpeg = true;
+                    out.push(0xFF);
+                    out.push(0xD8);
+                }
+                last = b;
+            } else {
+                out.push(b);
+                if last == 0xFF && b == 0xD9 {
+                    reader.consume(consumed);
+                    return Ok(out);
+                }
+                last = b;
+            }
+        }
+        reader.consume(consumed);
+    }
+}

sam-tool-tauri/src-tauri/src/extract.rs

@@ -0,0 +1,188 @@
+use crate::annotation::{AnnotationRecord, ImageAnnotations};
+use crate::coco::{Category, Shape};
+use crate::decoder::DecoderSession;
+use chrono::Utc;
+use std::path::Path;
+use uuid::Uuid;
+
+pub fn image_filename(idx: u64) -> String {
+    format!("f_{:04}.jpg", idx)
+}
+
+pub fn json_filename(idx: u64) -> String {
+    format!("f_{:04}.json", idx)
+}
+
+/// Parse a frame index from either the new `f_0042.jpg` pattern or the legacy
+/// `frame_000042.jpg` pattern. Returns None for unrelated files.
+fn parse_frame_idx(name: &str) -> Option<u64> {
+    let stem = name.strip_suffix(".jpg")?;
+    if let Some(rest) = stem.strip_prefix("f_") {
+        return rest.parse().ok();
+    }
+    if let Some(rest) = stem.strip_prefix("frame_") {
+        return rest.parse().ok();
+    }
+    None
+}
+
+pub fn extract_one(
+    decoder: &mut DecoderSession,
+    frame_idx: u64,
+    out_dir: &Path,
+    shapes: &[Shape],
+    author: &str,
+) -> Result<(), String> {
+    std::fs::create_dir_all(out_dir)
+        .map_err(|e| format!("mkdir {}: {e}", out_dir.display()))?;
+
+    // Snapshot info so we don't hold an immutable borrow across the mutable
+    // decode_at call below.
+    let info = decoder.info().clone();
+    write_folder_source(out_dir, &info.path);
+
+    let jpeg = decoder.decode_at(frame_idx)?;
+    let img_path = out_dir.join(image_filename(frame_idx));
+    std::fs::write(&img_path, &jpeg)
+        .map_err(|e| format!("write {}: {e}", img_path.display()))?;
+
+    let json_path = out_dir.join(json_filename(frame_idx));
+    if shapes.is_empty() {
+        // Clean up any stale sibling JSON from a prior extract with annotations.
+        let _ = std::fs::remove_file(&json_path);
+        return Ok(());
+    }
+
+    let now = Utc::now().to_rfc3339();
+    let annotations = shapes
+        .iter()
+        .map(|s| shape_to_record(s, author, &now))
+        .collect();
+
+    let ia = ImageAnnotations {
+        version: 1,
+        image: image_filename(frame_idx),
+        width: info.width,
+        height: info.height,
+        source_video: Some(info.path),
+        annotations,
+    };
+    let text = serde_json::to_string_pretty(&ia)
+        .map_err(|e| format!("serialize json: {e}"))?;
+    std::fs::write(&json_path, text)
+        .map_err(|e| format!("write {}: {e}", json_path.display()))?;
+    Ok(())
+}
+
+pub const FOLDER_SOURCE_FILENAME: &str = "source.json";
+
+#[derive(serde::Serialize, serde::Deserialize)]
+pub struct FolderSource {
+    pub source_video: String,
+    pub created_at: String,
+}
+
+/// Write (once, if missing) a folder-level metadata file recording which
+/// video this output folder was extracted from. Silent on failure — this is
+/// a best-effort breadcrumb, not a critical write.
+pub fn write_folder_source(out_dir: &Path, video_path: &str) {
+    let meta_path = out_dir.join(FOLDER_SOURCE_FILENAME);
+    if meta_path.exists() {
+        return;
+    }
+    let meta = FolderSource {
+        source_video: video_path.to_string(),
+        created_at: Utc::now().to_rfc3339(),
+    };
+    if let Ok(text) = serde_json::to_string_pretty(&meta) {
+        let _ = std::fs::write(&meta_path, text);
+    }
+}
+
+pub fn read_folder_source(out_dir: &Path) -> Option<FolderSource> {
+    let meta_path = out_dir.join(FOLDER_SOURCE_FILENAME);
+    let text = std::fs::read_to_string(&meta_path).ok()?;
+    serde_json::from_str(&text).ok()
+}
+
+pub fn undo_one(frame_idx: u64, out_dir: &Path) -> Result<(), String> {
+    // Try both naming schemes so legacy extracts are also cleanable.
+    for pair in [
+        (image_filename(frame_idx), json_filename(frame_idx)),
+        (format!("frame_{:06}.jpg", frame_idx), format!("frame_{:06}.json", frame_idx)),
+    ] {
+        let img = out_dir.join(&pair.0);
+        let json = out_dir.join(&pair.1);
+        if img.exists() {
+            std::fs::remove_file(&img).map_err(|e| format!("remove {}: {e}", img.display()))?;
+        }
+        if json.exists() {
+            std::fs::remove_file(&json).map_err(|e| format!("remove {}: {e}", json.display()))?;
+        }
+    }
+    Ok(())
+}
+
+pub fn list_extracted(out_dir: &Path) -> Result<Vec<u64>, String> {
+    if !out_dir.exists() {
+        return Ok(vec![]);
+    }
+    let mut out = Vec::new();
+    let entries = std::fs::read_dir(out_dir)
+        .map_err(|e| format!("read_dir {}: {e}", out_dir.display()))?;
+    for entry in entries {
+        let entry = entry.map_err(|e| format!("dirent: {e}"))?;
+        if let Some(name) = entry.file_name().to_str() {
+            if let Some(n) = parse_frame_idx(name) {
+                out.push(n);
+            }
+        }
+    }
+    out.sort();
+    Ok(out)
+}
+
+pub fn write_classes_txt(out_dir: &Path, categories: &[Category]) -> Result<(), String> {
+    std::fs::create_dir_all(out_dir)
+        .map_err(|e| format!("mkdir {}: {e}", out_dir.display()))?;
+    let mut sorted = categories.to_vec();
+    sorted.sort_by_key(|c| c.id);
+    let mut text = String::new();
+    for c in &sorted {
+        text.push_str(&c.name);
+        text.push('\n');
+    }
+    let path = out_dir.join("classes.txt");
+    std::fs::write(&path, text).map_err(|e| format!("write {}: {e}", path.display()))
+}
+
+fn shape_to_record(shape: &Shape, author: &str, ts: &str) -> AnnotationRecord {
+    match shape {
+        Shape::Bbox {
+            class_id,
+            class_name,
+            bbox,
+        } => AnnotationRecord::Bbox {
+            id: Uuid::new_v4().to_string(),
+            class_id: *class_id,
+            class_name: class_name.clone(),
+            bbox: *bbox,
+            author: author.to_string(),
+            created_at: ts.to_string(),
+            updated_at: ts.to_string(),
+        },
+        Shape::Polygon {
+            class_id,
+            class_name,
+            points,
+        } => AnnotationRecord::Polygon {
+            id: Uuid::new_v4().to_string(),
+            class_id: *class_id,
+            class_name: class_name.clone(),
+            points: points.clone(),
+            author: author.to_string(),
+            created_at: ts.to_string(),
+            updated_at: ts.to_string(),
+        },
+    }
+}

sam-tool-tauri/src-tauri/src/lib.rs

@@ -0,0 +1,307 @@
+mod annotate;
+mod annotation;
+mod coco;
+mod decoder;
+mod extract;
+mod sam;
+mod settings;
+mod state;
+mod user;
+mod video;
+
+use decoder::DecoderSession;
+use serde::Serialize;
+use state::AppState;
+use std::collections::HashMap;
+use std::path::{Path, PathBuf};
+use tauri::{ipc::Response, AppHandle, Emitter, State};
+use video::VideoInfo;
+
+#[tauri::command]
+fn ping() -> String {
+    "pong".into()
+}
+
+#[tauri::command]
+fn open_video(state: State<'_, AppState>, path: String) -> Result<VideoInfo, String> {
+    let p = PathBuf::from(&path);
+    if !p.exists() {
+        return Err(format!("file not found: {}", p.display()));
+    }
+    let info = video::probe(&p)?;
+    let mut guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+    *guard = Some(DecoderSession::open(p, info.clone()));
+    Ok(info)
+}
+
+#[tauri::command]
+fn close_video(state: State<'_, AppState>) -> Result<(), String> {
+    let mut guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+    *guard = None;
+    Ok(())
+}
+
+#[tauri::command]
+fn current_video(state: State<'_, AppState>) -> Result<Option<VideoInfo>, String> {
+    let guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+    Ok(guard.as_ref().map(|d| d.info().clone()))
+}
+
+#[tauri::command]
+fn decode_frame(state: State<'_, AppState>, frame_idx: u64) -> Result<Response, String> {
+    let bytes = {
+        let mut guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+        let decoder = guard.as_mut().ok_or("no video loaded")?;
+        decoder.decode_at(frame_idx)?
+    };
+    Ok(Response::new(bytes))
+}
+
+#[tauri::command]
+fn import_coco(path: String) -> Result<coco::ImportedCoco, String> {
+    let p = Path::new(&path);
+    if !p.exists() {
+        return Err(format!("file not found: {}", p.display()));
+    }
+    coco::import(p)
+}
+
+#[tauri::command]
+fn set_cache_capacity(state: State<'_, AppState>, capacity: usize) -> Result<usize, String> {
+    let clamped = capacity.clamp(decoder::MIN_CACHE_CAPACITY, decoder::MAX_CACHE_CAPACITY);
+    let guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+    if let Some(d) = guard.as_ref() {
+        d.set_cache_capacity(clamped);
+    }
+    Ok(clamped)
+}
+
+#[tauri::command]
+fn extract_frame(
+    state: State<'_, AppState>,
+    frame_idx: u64,
+    out_dir: String,
+    shapes: Vec<coco::Shape>,
+    author: String,
+) -> Result<(), String> {
+    let mut guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+    let decoder = guard.as_mut().ok_or("no video loaded")?;
+    extract::extract_one(decoder, frame_idx, Path::new(&out_dir), &shapes, &author)
+}
+
+#[tauri::command]
+fn undo_extract(frame_idx: u64, out_dir: String) -> Result<(), String> {
+    extract::undo_one(frame_idx, Path::new(&out_dir))
+}
+
+#[derive(Serialize, Clone)]
+struct BulkProgress {
+    done: u64,
+    total: u64,
+}
+
+#[tauri::command]
+fn bulk_extract(
+    state: State<'_, AppState>,
+    app: AppHandle,
+    start: u64,
+    end: u64,
+    out_dir: String,
+    shapes_by_frame: HashMap<String, Vec<coco::Shape>>,
+    author: String,
+) -> Result<u64, String> {
+    if end < start {
+        return Err("end must be >= start".into());
+    }
+    let mut guard = state.decoder.lock().map_err(|e| format!("lock: {e}"))?;
+    let decoder = guard.as_mut().ok_or("no video loaded")?;
+    let dir = PathBuf::from(&out_dir);
+    let total = end - start + 1;
+    let empty: Vec<coco::Shape> = Vec::new();
+    let mut done: u64 = 0;
+    for idx in start..=end {
+        let shapes = shapes_by_frame
+            .get(&idx.to_string())
+            .unwrap_or(&empty);
+        extract::extract_one(decoder, idx, &dir, shapes, &author)?;
+        done += 1;
+        if done == 1 || done % 5 == 0 || done == total {
+            let _ = app.emit("bulk-extract-progress", BulkProgress { done, total });
+        }
+    }
+    Ok(done)
+}
+
+#[tauri::command]
+fn bulk_undo(
+    app: AppHandle,
+    start: u64,
+    end: u64,
+    out_dir: String,
+) -> Result<u64, String> {
+    if end < start {
+        return Err("end must be >= start".into());
+    }
+    let dir = PathBuf::from(&out_dir);
+    let total = end - start + 1;
+    let mut done: u64 = 0;
+    for idx in start..=end {
+        extract::undo_one(idx, &dir)?;
+        done += 1;
+        if done == 1 || done % 20 == 0 || done == total {
+            let _ = app.emit("bulk-undo-progress", BulkProgress { done, total });
+        }
+    }
+    Ok(done)
+}
+
+#[tauri::command]
+fn list_extracted_frames(out_dir: String) -> Result<Vec<u64>, String> {
+    extract::list_extracted(Path::new(&out_dir))
+}
+
+#[tauri::command]
+fn write_classes_txt(out_dir: String, categories: Vec<coco::Category>) -> Result<(), String> {
+    extract::write_classes_txt(Path::new(&out_dir), &categories)
+}
+
+#[tauri::command]
+fn list_folder_images(folder: String) -> Result<Vec<annotate::ImageEntry>, String> {
+    annotate::list_images(Path::new(&folder))
+}
+
+#[tauri::command]
+fn read_image_bytes(path: String) -> Result<Response, String> {
+    let bytes = annotate::read_image_bytes(Path::new(&path))?;
+    Ok(Response::new(bytes))
+}
+
+#[tauri::command]
+fn load_image_annotations(
+    image_path: String,
+) -> Result<Option<annotation::ImageAnnotations>, String> {
+    annotate::load_image_annotations(Path::new(&image_path))
+}
+
+#[tauri::command]
+fn load_classes(folder: String) -> Result<Vec<String>, String> {
+    annotate::load_classes(Path::new(&folder))
+}
+
+#[tauri::command]
+fn append_class(folder: String, name: String) -> Result<Vec<String>, String> {
+    annotate::append_class(Path::new(&folder), &name)
+}
+
+#[tauri::command]
+fn delete_images(paths: Vec<String>) -> Result<u32, String> {
+    annotate::delete_images(&paths)
+}
+
+#[tauri::command]
+async fn sam_health_check(url: String) -> Result<bool, String> {
+    sam::health_check(&url).await
+}
+
+#[tauri::command]
+async fn sam_list_models(url: String) -> Result<sam::ModelList, String> {
+    sam::list_models(&url).await
+}
+
+#[tauri::command]
+async fn sam_segment_bbox(
+    url: String,
+    image_path: String,
+    bbox: [f64; 4],
+    image_width: u32,
+    image_height: u32,
+    model: Option<String>,
+) -> Result<Vec<[f64; 2]>, String> {
+    sam::segment_bbox(
+        &url,
+        Path::new(&image_path),
+        bbox,
+        image_width,
+        image_height,
+        model,
+    )
+    .await
+}
+
+#[tauri::command]
+fn save_image_annotations(
+    image_path: String,
+    payload: annotation::ImageAnnotations,
+) -> Result<(), String> {
+    let p = Path::new(&image_path);
+    let json_path = p.with_extension("json");
+    if payload.annotations.is_empty() {
+        if json_path.exists() {
+            std::fs::remove_file(&json_path)
+                .map_err(|e| format!("remove {}: {e}", json_path.display()))?;
+        }
+        return Ok(());
+    }
+    // Backfill source_video from folder-level source.json when the payload
+    // doesn't carry it — handles annotations authored via AnnotateMode that
+    // never went through extract_one.
+    let mut payload = payload;
+    if payload.source_video.is_none() {
+        if let Some(parent) = p.parent() {
+            if let Some(meta) = extract::read_folder_source(parent) {
+                payload.source_video = Some(meta.source_video);
+            }
+        }
+    }
+    let text = serde_json::to_string_pretty(&payload)
+        .map_err(|e| format!("serialize: {e}"))?;
+    std::fs::write(&json_path, text)
+        .map_err(|e| format!("write {}: {e}", json_path.display()))
+}
+
+#[tauri::command]
+fn read_folder_source(folder: String) -> Option<String> {
+    extract::read_folder_source(Path::new(&folder)).map(|s| s.source_video)
+}
+
+#[cfg_attr(mobile, tauri::mobile_entry_point)]
+pub fn run() {
+    tauri::Builder::default()
+        .plugin(tauri_plugin_dialog::init())
+        .plugin(tauri_plugin_fs::init())
+        .manage(AppState::new())
+        .invoke_handler(tauri::generate_handler![
+            ping,
+            user::current_user,
+            user::list_known_users,
+            user::set_current_user,
+            user::clear_current_user,
+            open_video,
+            close_video,
+            current_video,
+            decode_frame,
+            set_cache_capacity,
+            import_coco,
+            extract_frame,
+            undo_extract,
+            bulk_extract,
+            bulk_undo,
+            list_extracted_frames,
+            write_classes_txt,
+            list_folder_images,
+            read_image_bytes,
+            load_image_annotations,
+            load_classes,
+            save_image_annotations,
+            append_class,
+            delete_images,
+            read_folder_source,
+            settings::get_settings,
+            settings::save_settings,
+            sam_health_check,
+            sam_list_models,
+            sam_segment_bbox,
+        ])
+        .run(tauri::generate_context!())
+        .expect("error while running tauri application");
+}

sam-tool-tauri/src-tauri/src/main.rs

@@ -0,0 +1,6 @@
+// Prevents a console window on Windows release builds (harmless on Linux).
+#![cfg_attr(not(debug_assertions), windows_subsystem = "windows")]
+
+fn main() {
+    sam_tool_tauri_lib::run()
+}

sam-tool-tauri/src-tauri/src/sam.rs

@@ -0,0 +1,222 @@
+use base64::Engine as _;
+use serde::{Deserialize, Serialize};
+use serde_json::{json, Value};
+use std::path::Path;
+use std::time::Duration;
+
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct ModelInfo {
+    pub id: String,
+    pub name: String,
+    pub family: String,
+    pub available: bool,
+    #[serde(default)]
+    pub checkpoint: String,
+}
+
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct ModelList {
+    #[serde(default)]
+    pub default: Option<String>,
+    pub models: Vec<ModelInfo>,
+}
+
+/// Fetch the list of models the backend advertises. The Tauri Settings
+/// modal uses this to populate a dropdown so users can pick which variant
+/// to hit.
+pub async fn list_models(url: &str) -> Result<ModelList, String> {
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(6))
+        .build()
+        .map_err(|e| format!("http client: {e}"))?;
+    let endpoint = format!("{}/models", url.trim_end_matches('/'));
+    let res = client
+        .get(&endpoint)
+        .send()
+        .await
+        .map_err(|e| format!("connect to {endpoint}: {e}"))?;
+    if !res.status().is_success() {
+        return Err(format!("GET /models → {}", res.status()));
+    }
+    res.json::<ModelList>()
+        .await
+        .map_err(|e| format!("parse /models: {e}"))
+}
+
+/// Minimal XML that makes a Label-Studio-ML SAM backend happy. The names
+/// (`bbox`, `polygon`, `tag`) are referenced from the request / response by
+/// their `from_name` / `to_name` fields — if your backend expects different
+/// names we can wire them up.
+const LABEL_CONFIG: &str = r##"<View>
+  <Image name="image" value="$image" zoom="true"/>
+  <BrushLabels name="tag" toName="image">
+    <Label value="Object" background="#ff0000"/>
+  </BrushLabels>
+  <RectangleLabels name="bbox" toName="image" smart="true">
+    <Label value="Object" background="#ff0000"/>
+  </RectangleLabels>
+  <PolygonLabels name="polygon" toName="image" smart="true">
+    <Label value="Object" background="#ff0000"/>
+  </PolygonLabels>
+</View>"##;
+
+/// Ping the SAM backend's root.
+pub async fn health_check(url: &str) -> Result<bool, String> {
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(4))
+        .build()
+        .map_err(|e| format!("http client: {e}"))?;
+    let res = client
+        .get(url)
+        .send()
+        .await
+        .map_err(|e| format!("connect: {e}"))?;
+    Ok(res.status().is_success() || res.status().is_client_error())
+}
+
+/// POST the image + bbox prompt to a Label-Studio-ML SAM backend, return the
+/// polygon in **image-pixel** coords.
+///
+/// Wire protocol (LS-ML):
+/// ```json
+/// {
+///   "tasks": [{ "id": 1, "data": { "image": "data:image/jpeg;base64,..." } }],
+///   "label_config": "<View>...</View>",
+///   "params": {
+///     "context": {
+///       "result": [{
+///         "original_width": W, "original_height": H, "image_rotation": 0,
+///         "value": { "x": %, "y": %, "width": %, "height": %,
+///                    "rotation": 0, "rectanglelabels": ["Object"] },
+///         "id": "box_0",
+///         "from_name": "bbox", "to_name": "image",
+///         "type": "rectanglelabels"
+///       }]
+///     }
+///   }
+/// }
+/// ```
+pub async fn segment_bbox(
+    url: &str,
+    image_path: &Path,
+    bbox: [f64; 4],
+    image_width: u32,
+    image_height: u32,
+    model: Option<String>,
+) -> Result<Vec<[f64; 2]>, String> {
+    let image_bytes = std::fs::read(image_path)
+        .map_err(|e| format!("read image {}: {e}", image_path.display()))?;
+    let image_b64 = base64::engine::general_purpose::STANDARD.encode(&image_bytes);
+    let data_url = format!("data:image/jpeg;base64,{}", image_b64);
+
+    let [x, y, w, h] = bbox;
+    let width_f = image_width.max(1) as f64;
+    let height_f = image_height.max(1) as f64;
+    let x_pct = (x / width_f) * 100.0;
+    let y_pct = (y / height_f) * 100.0;
+    let w_pct = (w / width_f) * 100.0;
+    let h_pct = (h / height_f) * 100.0;
+
+    let body = json!({
+        "tasks": [{
+            "id": 1,
+            "data": { "image": data_url }
+        }],
+        "label_config": LABEL_CONFIG,
+        "params": {
+            "context": {
+                "result": [{
+                    "original_width": image_width,
+                    "original_height": image_height,
+                    "image_rotation": 0,
+                    "value": {
+                        "x": x_pct,
+                        "y": y_pct,
+                        "width": w_pct,
+                        "height": h_pct,
+                        "rotation": 0,
+                        "rectanglelabels": ["Object"]
+                    },
+                    "id": "box_0",
+                    "from_name": "bbox",
+                    "to_name": "image",
+                    "type": "rectanglelabels"
+                }]
+            }
+        }
+    });
+
+    let client = reqwest::Client::builder()
+        .timeout(Duration::from_secs(60))
+        .build()
+        .map_err(|e| format!("http client: {e}"))?;
+
+    let endpoint = format!("{}/predict", url.trim_end_matches('/'));
+    let mut req = client.post(&endpoint);
+    if let Some(ref m) = model {
+        if !m.is_empty() {
+            req = req.query(&[("model", m.as_str())]);
+        }
+    }
+    let res = req
+        .json(&body)
+        .send()
+        .await
+        .map_err(|e| format!("request to {endpoint}: {e}"))?;
+
+    if !res.status().is_success() {
+        let status = res.status();
+        let body_text = res.text().await.unwrap_or_default();
+        return Err(format!("SAM {} — {}", status, body_text));
+    }
+    let raw: Value = res
+        .json()
+        .await
+        .map_err(|e| format!("parse SAM response: {e}"))?;
+
+    extract_polygon(&raw, width_f, height_f)
+        .ok_or_else(|| format!("unrecognized SAM response shape: {}", raw))
+}
+
+/// Walk the JSON tree looking for a polygon — LS-ML usually returns
+/// `{ results: [{ result: [{ value: { points: [[x%, y%], ...] } }] }] }`,
+/// but we also accept plain `{ polygon: [[x, y], ...] }` in pixels.
+fn extract_polygon(v: &Value, w: f64, h: f64) -> Option<Vec<[f64; 2]>> {
+    // Plain shape: pixel polygon at the top level.
+    if let Some(poly) = v.get("polygon").and_then(|p| p.as_array()) {
+        let out = parse_points(poly, 1.0, 1.0);
+        if out.len() >= 3 {
+            return Some(out);
+        }
+    }
+
+    // LS-ML shape.
+    let results = v.get("results").and_then(|r| r.as_array())?;
+    let first = results.first()?;
+    let inner = first.get("result").and_then(|r| r.as_array())?;
+    for entry in inner {
+        let value = entry.get("value")?;
+        if let Some(points) = value.get("points").and_then(|p| p.as_array()) {
+            // LS polygons are in percentages of the original image size.
+            let out = parse_points(points, w / 100.0, h / 100.0);
+            if out.len() >= 3 {
+                return Some(out);
+            }
+        }
+    }
+    None
+}
+
+fn parse_points(arr: &[Value], sx: f64, sy: f64) -> Vec<[f64; 2]> {
+    let mut out = Vec::with_capacity(arr.len());
+    for pt in arr {
+        if let Some(p) = pt.as_array() {
+            if p.len() >= 2 {
+                if let (Some(x), Some(y)) = (p[0].as_f64(), p[1].as_f64()) {
+                    out.push([x * sx, y * sy]);
+                }
+            }
+        }
+    }
+    out
+}

sam-tool-tauri/src-tauri/src/settings.rs

@@ -0,0 +1,71 @@
+use serde::{Deserialize, Serialize};
+use std::fs;
+use std::path::PathBuf;
+use tauri::{AppHandle, Manager};
+
+#[derive(Serialize, Deserialize, Clone)]
+pub struct Settings {
+    #[serde(default = "default_sam_url")]
+    pub sam_url: String,
+    #[serde(default)]
+    pub sam_enabled: bool,
+    /// Model id chosen from the backend's /models list (e.g.
+    /// "sam2.1_hiera_small"). `None` lets the backend pick its default.
+    #[serde(default)]
+    pub sam_model: Option<String>,
+    #[serde(default = "default_cache_capacity")]
+    pub cache_capacity: usize,
+}
+
+fn default_sam_url() -> String {
+    "http://localhost:9090".into()
+}
+
+fn default_cache_capacity() -> usize {
+    500
+}
+
+impl Default for Settings {
+    fn default() -> Self {
+        Self {
+            sam_url: default_sam_url(),
+            sam_enabled: false,
+            sam_model: None,
+            cache_capacity: default_cache_capacity(),
+        }
+    }
+}
+
+fn settings_path(app: &AppHandle) -> Result<PathBuf, String> {
+    let dir = app
+        .path()
+        .app_local_data_dir()
+        .map_err(|e| format!("app data dir: {e}"))?;
+    fs::create_dir_all(&dir).map_err(|e| format!("mkdir {}: {e}", dir.display()))?;
+    Ok(dir.join("settings.json"))
+}
+
+pub fn load(app: &AppHandle) -> Result<Settings, String> {
+    let path = settings_path(app)?;
+    if !path.exists() {
+        return Ok(Settings::default());
+    }
+    let text = fs::read_to_string(&path).map_err(|e| format!("read {}: {e}", path.display()))?;
+    serde_json::from_str(&text).map_err(|e| format!("parse settings: {e}"))
+}
+
+pub fn save(app: &AppHandle, s: &Settings) -> Result<(), String> {
+    let path = settings_path(app)?;
+    let text = serde_json::to_string_pretty(s).map_err(|e| format!("serialize: {e}"))?;
+    fs::write(&path, text).map_err(|e| format!("write {}: {e}", path.display()))
+}
+
+#[tauri::command]
+pub fn get_settings(app: AppHandle) -> Result<Settings, String> {
+    load(&app)
+}
+
+#[tauri::command]
+pub fn save_settings(app: AppHandle, settings: Settings) -> Result<(), String> {
+    save(&app, &settings)
+}

sam-tool-tauri/src-tauri/src/state.rs

@@ -0,0 +1,14 @@
+use crate::decoder::DecoderSession;
+use std::sync::Mutex;
+
+pub struct AppState {
+    pub decoder: Mutex<Option<DecoderSession>>,
+}
+
+impl AppState {
+    pub fn new() -> Self {
+        Self {
+            decoder: Mutex::new(None),
+        }
+    }
+}

sam-tool-tauri/src-tauri/src/user.rs

@@ -0,0 +1,76 @@
+use serde::{Deserialize, Serialize};
+use std::fs;
+use std::path::PathBuf;
+use tauri::{AppHandle, Manager};
+
+#[derive(Default, Serialize, Deserialize)]
+struct UserStore {
+    current_user: Option<String>,
+    #[serde(default)]
+    known_users: Vec<String>,
+}
+
+fn store_path(app: &AppHandle) -> Result<PathBuf, String> {
+    let dir = app
+        .path()
+        .app_local_data_dir()
+        .map_err(|e| format!("could not resolve app data dir: {e}"))?;
+    fs::create_dir_all(&dir).map_err(|e| format!("mkdir {}: {e}", dir.display()))?;
+    Ok(dir.join("users.json"))
+}
+
+fn load(app: &AppHandle) -> Result<UserStore, String> {
+    let path = store_path(app)?;
+    if !path.exists() {
+        return Ok(UserStore::default());
+    }
+    let text = fs::read_to_string(&path).map_err(|e| format!("read {}: {e}", path.display()))?;
+    serde_json::from_str(&text).map_err(|e| format!("parse users.json: {e}"))
+}
+
+fn save(app: &AppHandle, store: &UserStore) -> Result<(), String> {
+    let path = store_path(app)?;
+    let text = serde_json::to_string_pretty(store).map_err(|e| format!("serialize: {e}"))?;
+    fs::write(&path, text).map_err(|e| format!("write {}: {e}", path.display()))
+}
+
+fn normalize(name: &str) -> Result<String, String> {
+    let trimmed = name.trim();
+    if trimmed.is_empty() {
+        return Err("username cannot be empty".into());
+    }
+    if trimmed.len() > 64 {
+        return Err("username too long (max 64 chars)".into());
+    }
+    Ok(trimmed.to_string())
+}
+
+#[tauri::command]
+pub fn current_user(app: AppHandle) -> Result<Option<String>, String> {
+    Ok(load(&app)?.current_user)
+}
+
+#[tauri::command]
+pub fn list_known_users(app: AppHandle) -> Result<Vec<String>, String> {
+    Ok(load(&app)?.known_users)
+}
+
+#[tauri::command]
+pub fn set_current_user(app: AppHandle, name: String) -> Result<String, String> {
+    let name = normalize(&name)?;
+    let mut store = load(&app)?;
+    if !store.known_users.iter().any(|u| u == &name) {
+        store.known_users.push(name.clone());
+        store.known_users.sort();
+    }
+    store.current_user = Some(name.clone());
+    save(&app, &store)?;
+    Ok(name)
+}
+
+#[tauri::command]
+pub fn clear_current_user(app: AppHandle) -> Result<(), String> {
+    let mut store = load(&app)?;
+    store.current_user = None;
+    save(&app, &store)
+}

sam-tool-tauri/src-tauri/src/video.rs

@@ -0,0 +1,106 @@
+use serde::Serialize;
+use std::path::Path;
+use std::process::Command;
+
+#[derive(Serialize, Clone, Debug)]
+pub struct VideoInfo {
+    pub path: String,
+    pub width: u32,
+    pub height: u32,
+    pub fps: f64,
+    pub total_frames: u64,
+    pub duration_s: f64,
+    /// Suggested output folder for this video, derived from the video path so
+    /// the same folder is produced on each re-open. Frontend uses this to
+    /// auto-restore the extracted-frame set without the user re-picking.
+    pub default_output_folder: String,
+}
+
+pub fn default_output_folder(path: &Path) -> String {
+    let stem = path
+        .file_stem()
+        .and_then(|s| s.to_str())
+        .unwrap_or("video");
+    let parent = path.parent().unwrap_or_else(|| Path::new("."));
+    parent
+        .join(format!("{stem}_extracted"))
+        .display()
+        .to_string()
+}
+
+fn parse_rational(s: &str) -> Option<f64> {
+    let (n, d) = s.split_once('/')?;
+    let n: f64 = n.parse().ok()?;
+    let d: f64 = d.parse().ok()?;
+    if d == 0.0 {
+        None
+    } else {
+        Some(n / d)
+    }
+}
+
+pub fn probe(path: &Path) -> Result<VideoInfo, String> {
+    let path_str = path.to_str().ok_or_else(|| "invalid path".to_string())?;
+    let out = Command::new("ffprobe")
+        .args([
+            "-v",
+            "error",
+            "-select_streams",
+            "v:0",
+            "-show_entries",
+            "stream=width,height,r_frame_rate,avg_frame_rate,nb_frames,duration",
+            "-show_entries",
+            "format=duration",
+            "-of",
+            "json",
+            path_str,
+        ])
+        .output()
+        .map_err(|e| format!("ffprobe spawn failed: {e}"))?;
+    if !out.status.success() {
+        return Err(format!(
+            "ffprobe exited {}: {}",
+            out.status,
+            String::from_utf8_lossy(&out.stderr)
+        ));
+    }
+    let json: serde_json::Value =
+        serde_json::from_slice(&out.stdout).map_err(|e| format!("ffprobe json: {e}"))?;
+    let stream = &json["streams"][0];
+    if stream.is_null() {
+        return Err("no video stream found".into());
+    }
+    let width = stream["width"]
+        .as_u64()
+        .ok_or("stream.width missing")? as u32;
+    let height = stream["height"]
+        .as_u64()
+        .ok_or("stream.height missing")? as u32;
+    let fps = stream["r_frame_rate"]
+        .as_str()
+        .and_then(parse_rational)
+        .or_else(|| stream["avg_frame_rate"].as_str().and_then(parse_rational))
+        .ok_or("cannot parse fps")?;
+    let duration_s = stream["duration"]
+        .as_str()
+        .and_then(|s| s.parse::<f64>().ok())
+        .or_else(|| {
+            json["format"]["duration"]
+                .as_str()
+                .and_then(|s| s.parse::<f64>().ok())
+        })
+        .ok_or("cannot parse duration")?;
+    let total_frames = stream["nb_frames"]
+        .as_str()
+        .and_then(|s| s.parse::<u64>().ok())
+        .unwrap_or_else(|| (fps * duration_s).round() as u64);
+    Ok(VideoInfo {
+        path: path.display().to_string(),
+        width,
+        height,
+        fps,
+        total_frames,
+        duration_s,
+        default_output_folder: default_output_folder(path),
+    })
+}

sam-tool-tauri/src-tauri/tauri.conf.json

@@ -0,0 +1,33 @@
+{
+  "$schema": "https://schema.tauri.app/config/2",
+  "productName": "SAM Tool",
+  "version": "0.1.0",
+  "identifier": "com.seekright.samtool",
+  "build": {
+    "beforeDevCommand": "npm run dev",
+    "devUrl": "http://localhost:1420",
+    "beforeBuildCommand": "npm run build",
+    "frontendDist": "../dist"
+  },
+  "app": {
+    "windows": [
+      {
+        "title": "SAM Tool",
+        "width": 1280,
+        "height": 860,
+        "resizable": true,
+        "fullscreen": false
+      }
+    ],
+    "security": {
+      "csp": null
+    }
+  },
+  "bundle": {
+    "active": false,
+    "targets": "all",
+    "icon": [
+      "icons/icon.png"
+    ]
+  }
+}