优化人像搜索 #1211
优化人像搜索 #1211
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1211 +/- ##
=======================================
Coverage 30.27% 30.27%
=======================================
Files 50 50
Lines 3121 3121
=======================================
Hits 945 945
Misses 2125 2125
Partials 51 51 ☔ View full report in Codecov by Sentry. |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
There was a problem hiding this comment.
PR Summary
This PR adds comprehensive face tagging functionality and text search capabilities to the face recognition demo. Here are the key changes:
- Added new
ImagePreview.vuecomponent for viewing and tagging detected faces with proper scaling, cropping and editing capabilities - Implemented automatic tag propagation system that applies tags to similar faces based on configurable similarity threshold (0.75)
- Added SQLite database support for face tags with new methods for tag management and text-based face searching
- Enhanced frontend with text search input and face tag visualization in both management and search views
- Added proper error handling and loading states for tag operations and search functionality
The implementation is well-structured but should consider adding database migration code for existing installations.
💡 (1/5) You can manually trigger the bot by mentioning @greptileai in a comment!
6 file(s) reviewed, 12 comment(s)
Edit PR Review Bot Settings | Greptile
| // 应缩放比例到坐标 | ||
| const x = face.face_coords.x1 * scaleX | ||
| const y = face.face_coords.y1 * scaleY | ||
| const w = (face.face_coords.x2 - face.face_coords.x1) * scaleX | ||
| const h = (face.face_coords.y2 - face.face_coords.y1) * scaleY |
There was a problem hiding this comment.
logic: Missing null check for face_coords before accessing x1/y1/x2/y2 properties. Could cause runtime errors if face detection fails.
| inputPlaceholder: '请输入标签名称' | ||
| }) | ||
|
|
||
| if (tag !== null) { |
There was a problem hiding this comment.
logic: tag !== null check will not catch empty strings. Should validate tag has meaningful content before making API call.
| const response = await fetch('/api/face/tag', { | ||
| method: 'POST', | ||
| headers: { | ||
| 'Content-Type': 'application/json', | ||
| }, | ||
| body: JSON.stringify(requestData) | ||
| }) |
There was a problem hiding this comment.
style: API call lacks error handling for network failures and timeout scenarios
| if (error !== 'cancel') { | ||
| ElMessage.error(error.message || '标签更新失败') | ||
| } |
There was a problem hiding this comment.
style: Error handling only checks for 'cancel' string, but ElMessageBox.prompt can reject with various values. Consider more robust error type checking.
| @@ -282,7 +283,7 @@ def delete_face_vectors(self, photo_id: int) -> bool: | |||
| # # 执行压缩操作以回收空间 | |||
| # self.collection.compact() | |||
|
|
|||
| logger.info(f"成功删除图片ID {photo_id} 的人脸向量") | |||
| logger.info(f"成功删除片ID {photo_id} 的人脸向量") | |||
There was a problem hiding this comment.
syntax: typo in log message: '片ID' should be '图片ID'
| similar_faces = db_manager.search_similar_faces(process_result[1]['embeddings'][i].tolist()) | ||
| if similar_faces: | ||
| # 找出相似度最大的人脸 | ||
| max_similarity_face = max(similar_faces, key=lambda face: face['similarity']) | ||
| if max_similarity_face['similarity'] > SIMILARITY_FOR_TAG: | ||
| tag = db_manager.get_face_tag(max_similarity_face['id'], max_similarity_face['face_index']) |
There was a problem hiding this comment.
logic: searching for similar faces before inserting the new face vectors could miss potential matches since the current face isn't in the database yet
| if similar_faces: | ||
| for face in similar_faces: | ||
| if db_manager.update_face_tag(face['id'], face['face_index'], request.tag): | ||
| updated_count += 1 | ||
|
|
||
| # 确保当前选中的人脸也被标记 | ||
| if db_manager.update_face_tag(request.image_id, request.face_index, request.tag): | ||
| updated_count += 1 |
There was a problem hiding this comment.
style: no transaction wrapping the multiple tag updates - could leave database in inconsistent state if operation fails partway through
| const handleTagViewerClose = () => { | ||
| showTagViewer.value = false | ||
| showViewer.value = false | ||
| previewUrl.value = '' | ||
| currentFaces.value = [] | ||
| } |
There was a problem hiding this comment.
logic: currentImage not being cleared in handleTagViewerClose
| .map(item => ({ | ||
| url: item.oss_url, | ||
| filename: item.original_filename, | ||
| similarity: Math.max(...item.faces.map(face => face.similarity)), |
There was a problem hiding this comment.
logic: same Math.max() issue with empty faces array as in image search
| const response = await axios.post('/api/text/search', { | ||
| query: searchText.value.trim() | ||
| }) |
There was a problem hiding this comment.
style: consider adding rate limiting or debouncing for text search to prevent API abuse
No description provided.