samuelstevens
commented
1 year ago 12 epochs later:
2022-11-07 23:26:49,546 - mmdet - INFO - Evaluating bbox...
Loading and preparing results...
DONE (t=0.86s)
creating index...
index created!
Running per image evaluation...
Evaluate annotation type *bbox*
DONE (t=43.22s).
Accumulating evaluation results...
DONE (t=9.55s).
2022-11-07 23:27:44,515 - mmdet - INFO -
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.350
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=1000 ] = 0.531
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=1000 ] = 0.411
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=1000 ] = 0.013
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=1000 ] = 0.245
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=1000 ] = 0.387
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.461
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=300 ] = 0.461
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=1000 ] = 0.461
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=1000 ] = 0.038
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=1000 ] = 0.354
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=1000 ] = 0.485
2022-11-07 23:27:45,429 - mmdet - INFO - Exp name: groovy_grape_fishnet.py
2022-11-07 23:27:45,430 - mmdet - INFO - Epoch(val) [12][2684] bbox_mAP: 0.3500, bbox_mAP_50: 0.5310, bbox_mAP_75: 0.4110, bbox_mAP_s: 0.0130, bbox_mAP_m: 0.2450, bbox_mAP_l: 0.3870, bbox_mAP_copypaste: 0.350 0.531 0.411 0.013 0.245 0.387
Dataset: Fishnet Data: 100% Model: Best iNat21 with hierarchical multitask pretraining
Use a pre-trained swin-v2-base model pre-trained on iNat21 with hierarchical multitask loss with the best validation loss, fine-tune it on Fishnet, then report the accuracy (and preferably attach training logs, link to WandB dashboards, etc).