Spaces:
Sleeping
Sleeping
Commit History
-
feat: add pose, general-objects, skin, brain, tabular endpoints
unknownModule 3 expansion backend — 7 new endpoints across 3 categories. Object Detection: - /detection/predict-pose: YOLOv8n-pose, 17 COCO keypoints, cap 10 persons - /detection/predict-general: YOLOv8n full 80 COCO classes Medical Imaging (unified multi-class schema with mode field + heatmap): - /medical/predict-skin-lesion: Anwarkh1/Skin_Cancer ViT (9 classes) - /medical/predict-brain-tumor: Devarshi/Brain_Tumor ViT (4 classes) - medical_cam.py: attention rollout for ViT + Grad-CAM for CNN Tabular (train-on-startup sklearn, per-request contributions): - /tabular/predict-titanic: LogisticRegression (coef-based contribution) - /tabular/predict-heart: GradientBoostingClassifier (LOCO approximation) - /tabular/predict-wine: GradientBoostingClassifier + quality bucketing - tabular_models.py: shared training + contribution logic - sample_data/heart.csv, wine_red.csv: UCI + Kaggle datasets Other: - /models/warmup: accept new keys (pose, general, skin, brain, titanic, heart, wine) - .gitignore…af3091e
-
feat: upgrade to EfficientNetV2-S + add torch/torchvision deps
unknownd5e772a