Upload notebooks/ASRM_matrix_both_datasets.ipynb with huggingface_hub

Upload code/Varroa_ASRM_training_code.zip with huggingface_hub

Upload code/Varroa_ASRM_training_code.zip with huggingface_hub

Upload notebooks/ASRM_matrix_both_datasets.ipynb with huggingface_hub

Upload notebooks/ASRM_matrix_both_datasets.ipynb with huggingface_hub

Upload code/Varroa_ASRM_training_code.zip with huggingface_hub

Upload code/Varroa_ASRM_training_code.zip with huggingface_hub

Upload notebooks/ASRM_matrix_both_datasets.ipynb with huggingface_hub

Load compatible pretrained backbone neck and Detect weights

Add ASRM candidate verification and sparse refinement case

Add context-aware ASRM P2/P3 refinement

Split P2+P3 into new sibling cases (p2p3_asrm_fusion, input_asrm_guided_p2p3), keep original P2-only cases intact

collect_results.py: add baseline_p2 + --cases to select a subset

Fix: V1+V3 apply ASRM to P2+P3 (V2 stays P2-only)

Sync READMEs (P2+P3 ASRM, unified multi-dataset notebook)

Add baseline_p2 case; ASRM V1/V2 now apply to P3 too

Add Varroa raw data and three-ASRM training workflow

initial commit