search

fizyr / keras-retinanet

Keras implementation of RetinaNet object detection.
Apache License 2.0
4.38k stars 1.96k forks source link

Ap value of each category in COCO dataset #1397

Closed mary-0830 closed 4 years ago

mary-0830 commented 4 years ago

Hello, I would like to ask, where can I find the code to calculate the ap value of each category?

My data set is similar to coco format, and currently only output such results: Average Precision (AP)@[ IoU=0. 50: 0. 95 area- all maxDets-=100 ]=0.384 Average Precision (AP)@[ IoU=0. 50area= all maxDets=100]=0.633 Average Precision(AP)@L IoU=0.75 area = all maxDets=100]=0.408 Average Precision (AP)@[ IoU=0.50: 0.95 area= small maxDets=100 Average Precision (AP)@[ IoU=0. 50: 0.95 area=medium maxDets=100]=0.408 Average Precision (AP)@[ IoU=0. 50: 0. 95 area= large max Dets=100]=0.479 Average Recal(AR)@[ IoU=0. 50:0.95 area= all maxDets=1]=0.214 Average Recall (AR)@[ IoU=0. 50:0.95 area= all maxDets-=10]=0.444 Average Recal(AR)@[ IoU=0. 50:0.95 area= all maxDets-=100]=0.543 Average Recall (AR)@[ IoU=0. 50:0.95 area= small maxDets=100]=0.305 Average Recall(AR)@[ IoU=0. 50:0.95 area-medium maxDets=100]=0.596 Average Recall(AR)Iou=0. 50:0.95 area= large maxDets=100]=0.649

I want to generate ap values for each category, please give me pointers. Thanks, looking forward to your reply.

Gateway2745 commented 4 years ago

It is in utils/eval.py https://github.com/fizyr/keras-retinanet/blob/d9065ef63ae16c4ef3dac00123ce38c07b8128a0/keras_retinanet/utils/eval.py#L30-L56

mary-0830 commented 4 years ago

It is in utils/eval.py https://github.com/fizyr/keras-retinanet/blob/d9065ef63ae16c4ef3dac00123ce38c07b8128a0/keras_retinanet/utils/eval.py#L30-L56

my cmd： python evaluate.py --image-min-side=600 --image-max-side=600 coco /home/zxc/DOTA/DOTA_clip_coco_600/DOTA_clip_coco_600/ ./models/resnet50_coco_06.h5

I still don’t understand.So what is my order changed?

stale[bot] commented 4 years ago

This issue has been automatically marked as stale due to the lack of recent activity. It will be closed if no further activity occurs. Thank you for your contributions.