Training of Faster R-CNN R50 on Detectron2 Pytorch

[wiky231]

Hi, I'm trying to train a Faster RCNN R50 model through Detectron2 Pytorch on Google Colab. My training dataset consists of about 5000 images and I'm training for 45000 iterations with a batch size of 16. The model is evaluated on the validation set every 500 iterations and I get the info attached below, can I know which of the values below represents mAP@0.5? And what does the APm stand for? Thanks in advance.

| AP | AP50 | AP75 | APs | APm | APl | |:------:|:------:|:------:|:------:|:------:|:------:| | 23.208 | 38.640 | 24.023 | 10.269 | 22.183 | 25.567 | [05/12 03:46:13 d2.evaluation.coco_evaluation]: Per-category bbox AP: | category | AP | category | AP | category | AP | |:----------------|:-------|:-------------|:-------|:-------------------|:-------| | Marine-debris | nan | animal_crab | 22.216 | animal_eel | 26.423 | | animal_etc | 0.886 | animal_fish | 37.697 | animal_shells | 9.109 | | animal_starfish | 0.180 | plant | 23.591 | rov | 46.928 | | trash_etc | 17.811 | trash_fabric | 15.715 | trash_fishing_gear | 0.000 | | trash_metal | 35.573 | trash_paper | 29.547 | trash_plastic | 30.209 | | trash_rubber | 50.202 | trash_wood | 25.238 | | |

[AliNaqvi110]

Have you figure out answer kindly share it