feat: 完善视频传播、标注编辑和拆帧闭环
- 接入 SAM2 视频传播能力：新增 /api/ai/propagate，支持用当前帧 mask/polygon/bbox 作为 seed，通过 SAM2 video predictor 向前、向后或双向传播，并可保存为真实 annotation。 - 接入 SAM3 video tracker：通过独立 Python 3.12 external worker 调用 SAM3 video predictor/tracker，使用本地 checkpoint 与 bbox seed 执行视频级跟踪，并在模型状态中标记 video_track 能力。 - 完善 SAM 模型分发：sam_registry 按 model_id 明确区分 sam2 propagation 与 sam3 video_track，避免两个模型链路混用。 - 打通前端“传播片段”：VideoWorkspace 使用当前选中 mask 和当前 AI 模型调用后端传播接口，传播结果回写并刷新工作区已保存标注。 - 增强 SAM3 本地 checkpoint 配置：新增 sam3_checkpoint_path 配置和 .env.example 示例，状态检查改为基于本地 checkpoint/独立环境/模型包可用性。 - 完善视频拆帧参数：/api/media/parse 支持 parse_fps、max_frames、target_width，后端任务保存帧时间戳、源帧号和 frame_sequence 元数据。 - 增加运行时 schema 兼容处理：启动时为旧 frames 表补充 timestamp_ms 和 source_frame_number 列，避免旧库升级后缺字段。 - 强化 Canvas 标注编辑：补齐多边形闭合、点工具、顶点拖拽、边中点插入、Delete/Backspace 删除、区域合并和重叠去除等交互。 - 增强语义分类联动：选中 mask 后可通过右侧语义分类树更新标签、颜色和 class metadata，并同步到保存/导出链路。 - 增加关键帧时间轴体验：FrameTimeline 显示具体时间信息，并支持键盘左右方向键切换关键帧。 - 完善 AI 交互分割参数：前端保留正向点、反向点、框选和 interactive prompt 的调用状态，支持 SAM2 细化候选区域与 SAM3 bbox 入口。 - 扩展后端/前端 API 类型：新增 propagateMasks、传播请求/响应 schema，并补齐 annotation、导出、模型状态和任务接口的测试覆盖。 - 更新项目文档：同步 README、AGENTS、接口契约、需求冻结、设计冻结、前端元素审计、实施计划和测试计划，标明真实功能边界与剩余风险。 - 增加测试覆盖：补充 SAM2/SAM3 传播、SAM3 状态、媒体拆帧参数、Canvas 编辑、语义标签切换、时间轴、工作区传播和 API 合约测试。 - 加强仓库安全边界：将 sam3权重/ 加入 .gitignore，避免本地模型权重被误提交。 验证：npm run test:run；pytest backend/tests；npm run lint；npm run build；python -m py_compile；git diff --check。
This commit is contained in:
@@ -116,6 +116,44 @@ def test_predict_box_and_semantic_fallback(client, monkeypatch):
|
||||
assert semantic_response.json()["scores"] == [0.5]
|
||||
|
||||
|
||||
def test_predict_interactive_combines_box_and_points(client, monkeypatch):
|
||||
_, frame, _ = _create_project_and_frame(client)
|
||||
calls = {}
|
||||
monkeypatch.setattr("routers.ai._load_frame_image", lambda frame: np.zeros((10, 10, 3), dtype=np.uint8))
|
||||
|
||||
def fake_predict_interactive(model, image, box, points, labels):
|
||||
calls["model"] = model
|
||||
calls["box"] = box
|
||||
calls["points"] = points
|
||||
calls["labels"] = labels
|
||||
return (
|
||||
[[[0.2, 0.2], [0.8, 0.2], [0.8, 0.8]]],
|
||||
[0.88],
|
||||
)
|
||||
|
||||
monkeypatch.setattr("routers.ai.sam_registry.predict_interactive", fake_predict_interactive)
|
||||
|
||||
response = client.post("/api/ai/predict", json={
|
||||
"image_id": frame["id"],
|
||||
"prompt_type": "interactive",
|
||||
"prompt_data": {
|
||||
"box": [0.1, 0.1, 0.9, 0.9],
|
||||
"points": [[0.5, 0.5], [0.2, 0.2]],
|
||||
"labels": [1, 0],
|
||||
},
|
||||
"model": "sam2",
|
||||
})
|
||||
|
||||
assert response.status_code == 200
|
||||
assert response.json()["scores"] == [0.88]
|
||||
assert calls == {
|
||||
"model": "sam2",
|
||||
"box": [0.1, 0.1, 0.9, 0.9],
|
||||
"points": [[0.5, 0.5], [0.2, 0.2]],
|
||||
"labels": [1, 0],
|
||||
}
|
||||
|
||||
|
||||
def test_model_status_reports_runtime(client, monkeypatch):
|
||||
monkeypatch.setattr("routers.ai.sam_registry.runtime_status", lambda selected_model=None: {
|
||||
"selected_model": selected_model or "sam2",
|
||||
@@ -170,6 +208,80 @@ def test_model_status_reports_runtime(client, monkeypatch):
|
||||
assert body["models"][1]["available"] is False
|
||||
|
||||
|
||||
def test_propagate_saves_tracked_annotations(client, monkeypatch):
|
||||
project = client.post("/api/projects", json={"name": "Video Project"}).json()
|
||||
frames = [
|
||||
client.post(f"/api/projects/{project['id']}/frames", json={
|
||||
"project_id": project["id"],
|
||||
"frame_index": idx,
|
||||
"image_url": f"frames/{idx}.jpg",
|
||||
"width": 640,
|
||||
"height": 360,
|
||||
}).json()
|
||||
for idx in range(3)
|
||||
]
|
||||
calls = {}
|
||||
monkeypatch.setattr("routers.ai.download_file", lambda object_name: b"jpeg")
|
||||
|
||||
def fake_propagate_video(model, frame_paths, source_frame_index, seed, direction, max_frames):
|
||||
calls["model"] = model
|
||||
calls["source_frame_index"] = source_frame_index
|
||||
calls["seed"] = seed
|
||||
calls["direction"] = direction
|
||||
calls["max_frames"] = max_frames
|
||||
calls["frame_count"] = len(frame_paths)
|
||||
return [
|
||||
{
|
||||
"frame_index": 0,
|
||||
"polygons": [[[0.1, 0.1], [0.2, 0.1], [0.2, 0.2]]],
|
||||
"scores": [0.9],
|
||||
"object_ids": [1],
|
||||
},
|
||||
{
|
||||
"frame_index": 1,
|
||||
"polygons": [[[0.15, 0.15], [0.25, 0.15], [0.25, 0.25]]],
|
||||
"scores": [0.8],
|
||||
"object_ids": [1],
|
||||
},
|
||||
]
|
||||
|
||||
monkeypatch.setattr("routers.ai.sam_registry.propagate_video", fake_propagate_video)
|
||||
|
||||
response = client.post("/api/ai/propagate", json={
|
||||
"project_id": project["id"],
|
||||
"frame_id": frames[0]["id"],
|
||||
"model": "sam2",
|
||||
"direction": "forward",
|
||||
"max_frames": 2,
|
||||
"include_source": False,
|
||||
"seed": {
|
||||
"polygons": [[[0.1, 0.1], [0.2, 0.1], [0.2, 0.2]]],
|
||||
"bbox": [0.1, 0.1, 0.1, 0.1],
|
||||
"label": "胆囊",
|
||||
"color": "#ff0000",
|
||||
"class_metadata": {"id": "c1", "name": "胆囊", "color": "#ff0000", "zIndex": 20},
|
||||
"template_id": None,
|
||||
},
|
||||
})
|
||||
|
||||
assert response.status_code == 200
|
||||
body = response.json()
|
||||
assert body["created_annotation_count"] == 1
|
||||
assert body["processed_frame_count"] == 2
|
||||
assert calls["model"] == "sam2"
|
||||
assert calls["source_frame_index"] == 0
|
||||
assert calls["direction"] == "forward"
|
||||
assert calls["frame_count"] == 2
|
||||
saved = body["annotations"][0]
|
||||
assert saved["frame_id"] == frames[1]["id"]
|
||||
assert saved["mask_data"]["source"] == "sam2_propagation"
|
||||
assert saved["mask_data"]["class"]["name"] == "胆囊"
|
||||
assert saved["mask_data"]["score"] == 0.8
|
||||
|
||||
listing = client.get(f"/api/ai/annotations?project_id={project['id']}")
|
||||
assert len(listing.json()) == 1
|
||||
|
||||
|
||||
def test_predict_validation_errors(client, monkeypatch):
|
||||
project, _, _ = _create_project_and_frame(client)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user