Spaces:
Sleeping
Sleeping
Commit History
-
chore(forensics): finalize verification and trace validation\n\n- Fix unreachable background reasoning in inference.py\n- Enable enabling GPT reasoning by default in frontend\n- Verify correct data mapping for 15 naval fields\n- Clean up legacy code paths and ensure backward compatibility
Zhen Ye2fb805f
-
feat(threat-assessment): implement naval threat analysis with GPT-4o\n\n- Rename utils/gpt_distance.py to utils/gpt_reasoning.py and update logic for 15 naval threat features\n- Add Pydantic schemas for NavalThreatAssessment in utils/schemas.py\n- Update backend (app.py, inference.py) to use new threat estimation and pass full metadata\n- refactor(frontend): render threat level badges and detailed feature table in UI
Zhen Ye8094b21
-
Fix: IndexError in ByteTracker
Zhen Ye0834617 -
feat: replace SimpleTracker with ByteTrack
Zhen Ye6896025 -
Show first frame radar when processed image appears
Zhen Ye +1: Claude Opus 4.5- Add state.firstFrameReady flag set after first frame is processed - First frame radar now displays with processed image (not after video done) - Reset flag on video unload Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>85ec659
-
Add military-style radar with helicopter demo mode
Zhen Ye +1: Claude Opus 4.5- Redesign radar with military tactical aesthetic (green phosphor CRT look) - Add 4 helicopter tracks with keyframe interpolation for smooth movement - H01 moves away (triangle points up), H02-H04 approach (triangles point down) - First frame radar shows static snapshot after Reason clicked - Live radar shows animated view after Engage clicked - Remove old unused demo data files Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>d496e6d -
Refine overlays: Hide on first frame and remove speed from labels
Zhen Ye7a7588e
-
Respect CUDA_VISIBLE_DEVICES env var (do not delete it)
Zhen Yed3c7792 -
Fix NameError: import set_track_data in inference.py
Zhen Ye03a0884 -
Add writer loop error logging
Zhen Yeeb52404 -
Add aggressive worker debug logging and BaseException catch
Zhen Ye6a62982 -
Add granular logging for frame processing and ffmpeg
Zhen Yebbef397 -
Fix backend worker pipeline stall on exception (add fallback fallback)
Zhen Ye955fa9e -
Fix video playback race condition, local file protocol, and backend worker crash
Zhen Ye19fdabe
-
feat: Implement depth-based distance estimation and frontend integration
Zhen Ye78d352c -
fix: Increase queue buffer to absorb GPT latency
Zhen Ye89b854c -
feat: Radar historic traces and fix DETR crash
Zhen Yefe670b9 -
feat: Real-time radar directionality via track sync
Zhen Ye3a793fe -
fix: Import Full/Empty from queue to prevent NameError on backpressure
Zhen Ye90f0a6b -
fix: Switch to gpt-4o-mini and verified queue safety
Zhen Ye68bac29 -
fix: Update static file path for deployment
Zhen Ye22ef6b4 -
feat: Continuous object tracking, speed estimation, and overlay syncing
Zhen Yeff50694
-
Refactor frontend into modular ES modules in frontend/ folder
Zhen Yec17ec01 -
Fix tracking bugs: remove COCO fallback in HF mode, smooth velocity, and add track pruning
Zhen Ye811bd37 -
Use DETR no_timm revision to avoid timm dependency
Zhen Yed52f7cc -
Fix BGR to RGB conversion for DETR and GroundingDino inference
Zhen Yef78d96f -
Fix inference concurrency bugs and enable default first-frame GPT
Zhen Ye06e44d3 -
fix: Move CUDA_VISIBLE_DEVICES clearing to module import time
Zhen Ye- Moved env var clearing to very top of inference.py before torch import - Removed redundant debug logging and clearing logic from run_inference - Ensures all 4 GPUs are visible before PyTorch CUDA context initializes - Fixes HF Spaces issue where CUDA_VISIBLE_DEVICES=0 locks to single GPU9ea2cfe -
fix: Resolve IndentationError in DroneYoloDetector due to displaced import
Zhen Ye4e93b33 -
feat: Implement SAHI Tiling for 4K video detection
Zhen Ye- Added utils/tiling.py for image slicing and NMS - Updated DroneYolo and HFYoloV8 to auto-tile images > 3000px width - Uses 1280x1280 tiles with 20% overlap for maximum small object recallaf29397 -
optimize: Force high-res (1280px) inference for YOLO models to improve small object detection
Zhen Ye284ce20 -
perf: Tune batch sizes and queues for A10 GPUs
Zhen Ye- Increased YOLO (HF/Drone) batch size to 32 - Increased Depth Anything batch size to 16 - Increased SAM3 batch size to 8 - Increased DETR batch size to 8 - Expanded inference.py queue_out to handle 32-frame batch burstsa5f8d15 -
feat: Implement AsyncVideoReader for parallel video decoding
Zhen YeAddresses CPU bottleneck during inference by moving video decoding to a background thread with a prefetch buffer. Integrated into inference pipeline.356dce8
-
Simplify drone YOLO loading with ultralytics native HF Hub support
Zhen Ye2392266 -
Uncomment CUDA_VISIBLE_DEVICES clearing to enable all GPUs
Zhen Yee43c504 -
Update Lite model to use spencercdz/YOLOv8m_defence
Zhen Ye38a60a8 -
Remove InternVL2 logic as requested
Zhen Ye29e1c2b -
Fix internvl dependency subdirectory
Zhen Ye48d3b52 -
Fix internvl dependency github url
Zhen Ye9bee0f7 -
Add internvl dependency and disable frontend caching
Zhen Ye51193fd -
Fix SAM3 batch prediction shape mismatch and add InternVL2 to frontend
Zhen Ye3de3df3 -
feat: Integrate InternVL2 and fix SAM3 segmentation batch size issue
Zhen Ye3d32b4a -
Optimize streaming buffer and implement parallel depth pipeline
Zhen Ye8d938e9
-
Refactor: Decouple depth estimator and suppress noisy logs
Zhen Ye- Backend: Remove heuristic depth calculation; depth model now only outputs relative depth for visualization. - Frontend: Establish GPT as the sole authority for track distances. - Config: Reduce httpx/huggingface logging noise to WARNING level.64bbe44 -
Implement Batch Inference & Queue Backpressure Fixes
Zhen Ye5c36daa
-
fix: Handle empty GPT responses and log outputs
Zhen Ye1eea4fe -
feat: Implement Tab 2 GPT-based ranging and remove legacy depth calc
Zhen Yee16d22f -
Align Tab 2 Radar and UI with Tab 1
Zhen Ye6f9d10a -
Enable throttled fallback tracking for Tab 2 consistency
Zhen Ye6f9da27 -
Optimize MJPEG stream and fix first-frame UX
Zhen Ye16cab91