COCO Results

[haoruij677]

Hi, thanks for your implementation. Have you got the results from COCO yet? And I'm still wondering why its performance is worse than your faster-rcnn implementation?

[gurkirt]

Any update?

[gaoyao123]

I have trained this code in COCO2017 dataset.I used resnet101 for training and other parameters were invariant.I got the following results.Just for reference.

ROI Align
checkpoint 5
mAP = 0.25 mAP(Person)=0.380

AP
IOU area maxDets AP
0.50:0.95 all 100 0.250
0.50 all 100 0.445
0.75 all 100 0.258
0.50:0.95 small 100 0.115
0.50:0.95 medium 100 0.282
0.50:0.95 large 100 0.344

AR
IOU area maxDets AP
0.50:0.95 all 1 0.246
0.50:0.95 all 10 0.367
0.50:0.95 all 100 0.374
0.50:0.95 small 100 0.206
0.50:0.95 medium 100 0.406
0.50:0.95 large 100 0.521

[gurkirt]

Thank, I got mine working, see it at https://github.com/gurkirt/FPN.pytorch1.0
If you are looking for pytorch implementation

[gaoyao123]

OK,thanks a lot.

[CharlesPikachu]

Here is my result of resnet101 epoch 12 after updating this repo to torch1.x and borrow some codes from mmdet:

Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.374
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.587
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.406
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.208
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.417
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.487
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.313
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.493
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.517
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.319
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.562
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.664

code is here: https://github.com/DetectionBLWX/FPN.pytorch
I'm still working on this repo to obtain more reasonable results