Spaces:
Paused
Paused
Commit History
-
Fix AMP deprecation warnings
54f14c3 -
Fix ZeroGPU model loading lifecycle
337b262 -
Fix ZeroGPU startup and local GPU inference path
19d9b40
-
align project with CLAUDE spec and hf space deploy
cf54850
-
feat(fingerprint): improve CLIP attribution accuracy and add 4th detector
- Rewrite GENERATOR_PROMPTS to describe visual forensic artifacts instead of content style — CLIP picks up texture/frequency cues better than generator names - Add umm-maybe/AI-image-detector (EfficientNet fine-tuned on AI vs real) as 4th ensemble detector with weight 0.1; rebalance existing weights to [0.4,0.3,0.2,0.1] - Raise CLIP confidence threshold 0.25 → 0.32 (2.9× above chance for 9 classes) to reduce incorrect attributions when CLIP is uncertain39f9e8e -
feat: update generator labels and enhance detection response structure
1152d7b -
feat: Enhance generator detection and attribution mechanisms
- Introduced NoveltyDetector for detecting unseen generators using a CLIP embedding ring buffer and IsolationForest. - Added GeneratorRegistry for monitoring generator performance and retention, flagging those below a defined threshold. - Updated FingerprintEngine to include DCT frequency analysis and improved generator attribution logic. - Enhanced CoherenceEngine with audio lip-sync analysis, integrating it into the video processing pipeline. - Implemented Dempster-Shafer evidence fusion in Fuser for more robust verdicts based on multiple engine outputs. - Revised generator labels and their corresponding prompts to align with the updated taxonomy. - Added support for audio coherence sub-scores and timestamp markers in detection responses.8cf40cb
-
feat: add metadata extraction and synthetic keyword detection for images and videos
7250f83
-
refactor: clean up model roster comments and remove broken model references
10580aa -
fix: update Dockerfile dependencies, remove audio processing, and replace models
a38026a -
feat: enhance model capabilities and update dependencies
- Updated Dockerfile to include additional libraries for improved functionality. - Upgraded PyTorch and torchvision versions in requirements.txt for better performance. - Enhanced CLAUDE.md with user preferences and task execution guidelines. - Modified main.py to refine model inventory and weights for improved accuracy. - Improved error handling and logging in engine.py files for better debugging. - Added new audio processing capabilities and updated fingerprint engine logic.8363f67 -
Update model stack, add audio stream extraction, and harden inference wiring
1e24aab -
Initial commit
4e75170