Files
Pre_Seg_Server/backend/routers/media.py
admin 5ab4602535 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。
2026-05-01 20:27:33 +08:00

218 lines
7.3 KiB
Python

"""Media upload and parsing endpoints."""
import logging
from pathlib import Path
from typing import List, Optional
from fastapi import APIRouter, Depends, File, Form, HTTPException, Query, UploadFile, status
from sqlalchemy.orm import Session
from database import get_db
from minio_client import upload_file, get_presigned_url
from models import ProcessingTask, Project
from progress_events import publish_task_progress_event
from schemas import ProcessingTaskOut
from statuses import PROJECT_STATUS_PARSING, PROJECT_STATUS_PENDING, TASK_STATUS_QUEUED
from worker_tasks import parse_project_media
logger = logging.getLogger(__name__)
router = APIRouter(prefix="/api/media", tags=["Media"])
ALLOWED_EXTENSIONS = {".mp4", ".avi", ".mov", ".mkv", ".webm", ".png", ".jpg", ".jpeg", ".dcm"}
def _get_ext(filename: str) -> str:
return Path(filename).suffix.lower()
@router.post(
"/upload",
status_code=status.HTTP_201_CREATED,
summary="Upload a media file",
)
async def upload_media(
file: UploadFile = File(...),
project_id: Optional[int] = Form(None),
db: Session = Depends(get_db),
) -> dict:
"""Accept a video, image, or DICOM file and store it in MinIO.
If project_id is provided, the video_path of the project is updated.
Returns the presigned URL of the uploaded object.
"""
if not file.filename:
raise HTTPException(status_code=400, detail="Missing filename")
ext = _get_ext(file.filename)
if ext not in ALLOWED_EXTENSIONS:
raise HTTPException(
status_code=400,
detail=f"Unsupported file type: {ext}",
)
data = await file.read()
object_name = f"uploads/{project_id or 'general'}/{file.filename}"
try:
upload_file(object_name, data, content_type=file.content_type or "application/octet-stream", length=len(data))
except Exception as exc: # noqa: BLE001
logger.error("Upload failed: %s", exc)
raise HTTPException(status_code=500, detail="Upload to storage failed") from exc
file_url = get_presigned_url(object_name, expires=3600)
if project_id:
project = db.query(Project).filter(Project.id == project_id).first()
if project:
project.video_path = object_name
db.commit()
logger.info("Linked upload to project_id=%s", project_id)
else:
logger.warning("Project id=%s not found for upload linkage", project_id)
else:
# Auto-create a project named after the file
project = Project(
name=file.filename,
description="Auto-created from upload",
status=PROJECT_STATUS_PENDING,
video_path=object_name,
source_type="video",
)
db.add(project)
db.commit()
db.refresh(project)
project_id = project.id
object_name = f"uploads/{project_id}/{file.filename}"
# Re-upload with corrected path
upload_file(object_name, data, content_type=file.content_type or "application/octet-stream", length=len(data))
project.video_path = object_name
db.commit()
logger.info("Auto-created project id=%s for upload %s", project_id, file.filename)
logger.info("Upload complete: %s (size=%d bytes). Async parsing queued.", object_name, len(data))
return {
"object_name": object_name,
"file_url": file_url,
"size": len(data),
"project_id": project_id,
"message": "Upload successful. Parsing job queued.",
}
@router.post(
"/upload/dicom",
status_code=status.HTTP_201_CREATED,
summary="Upload multiple DICOM files",
)
async def upload_dicom_batch(
files: List[UploadFile] = File(...),
project_id: Optional[int] = Form(None),
db: Session = Depends(get_db),
) -> dict:
"""Upload multiple .dcm files for a DICOM series.
If project_id is provided, files are added to the existing project.
Otherwise a new DICOM project is created.
"""
if not files:
raise HTTPException(status_code=400, detail="No files uploaded")
uploaded = []
if project_id:
project = db.query(Project).filter(Project.id == project_id).first()
if not project:
raise HTTPException(status_code=404, detail="Project not found")
else:
# Create new DICOM project
first_name = files[0].filename or "DICOM_Series"
project = Project(
name=first_name,
description=f"DICOM series with {len(files)} files",
status=PROJECT_STATUS_PENDING,
source_type="dicom",
)
db.add(project)
db.commit()
db.refresh(project)
project_id = project.id
logger.info("Auto-created DICOM project id=%s", project_id)
for file in files:
if not file.filename or not file.filename.lower().endswith(".dcm"):
continue
data = await file.read()
object_name = f"uploads/{project_id}/dicom/{file.filename}"
try:
upload_file(object_name, data, content_type="application/dicom", length=len(data))
uploaded.append(object_name)
except Exception as exc: # noqa: BLE001
logger.error("Failed to upload DICOM %s: %s", file.filename, exc)
project.video_path = f"uploads/{project_id}/dicom"
db.commit()
return {
"project_id": project_id,
"uploaded_count": len(uploaded),
"message": f"Uploaded {len(uploaded)} DICOM files. Parsing job queued.",
}
@router.post(
"/parse",
status_code=status.HTTP_202_ACCEPTED,
response_model=ProcessingTaskOut,
summary="Trigger frame extraction",
)
def parse_media(
project_id: int,
source_type: Optional[str] = None,
parse_fps: Optional[float] = Query(None, gt=0, le=120),
max_frames: Optional[int] = Query(None, gt=0),
target_width: int = Query(640, ge=64, le=4096),
db: Session = Depends(get_db),
) -> ProcessingTask:
"""Create a background task for media frame extraction.
The Celery worker performs the heavy FFmpeg/OpenCV/pydicom work and
updates the persisted task record as it progresses.
"""
project = db.query(Project).filter(Project.id == project_id).first()
if not project:
raise HTTPException(status_code=404, detail="Project not found")
if not project.video_path:
raise HTTPException(status_code=400, detail="Project has no media uploaded")
effective_source = source_type or project.source_type or "video"
effective_parse_fps = parse_fps or project.parse_fps or 30.0
task = ProcessingTask(
task_type=f"parse_{effective_source}",
status=TASK_STATUS_QUEUED,
progress=0,
message="解析任务已入队",
project_id=project_id,
payload={
"source_type": effective_source,
"parse_fps": effective_parse_fps,
"max_frames": max_frames,
"target_width": target_width,
},
)
project.parse_fps = effective_parse_fps
project.status = PROJECT_STATUS_PARSING
db.add(task)
db.commit()
db.refresh(task)
publish_task_progress_event(task)
async_result = parse_project_media.delay(task.id)
task.celery_task_id = async_result.id
db.commit()
db.refresh(task)
logger.info("Queued parse task id=%s project_id=%s celery_id=%s", task.id, project_id, async_result.id)
return task