an error in eval.py about efficientdet.pytorch HOT 7 OPEN

hi，
if phase not in ['train', 'test'], it should run this following lines

list_transforms = []
list_transforms.extend([ albu.Normalize(mean=(0.485, 0.456, 0.406),std=(0.229, 0.224, 0.225), p=1), ToTensor() ])
albu.Compose(list_transforms, bbox_params=albu.BboxParams(format='pascal_voc', min_area=min_area, min_visibility=min_visibility, label_fields=['category_id']))

from efficientdet.pytorch.

I meet the same question, and I have printed out all the feature map sizes. The problem I think is miss of resizing the input image. I also don't know the solution, and I change the 'valid' to 'train' and 'test' and find that the mAp is very low(13% for train and 0.6% for test). Is there any solution to the problem we have met?

from efficientdet.pytorch.

hi，
if phase not in ['train', 'test'], it should run this following lines
list_transforms = []
list_transforms.extend([ albu.Normalize(mean=(0.485, 0.456, 0.406),std=(0.229, 0.224, 0.225), p=1), ToTensor() ])
albu.Compose(list_transforms, bbox_params=albu.BboxParams(format='pascal_voc', min_area=min_area, min_visibility=min_visibility, label_fields=['category_id']))

yes, but it throws a runtimeerror... I think maybe I should resize the images and then feed them to the net, just like the 'test' mode does.

from efficientdet.pytorch.

from efficientdet.pytorch.

I meet the same question, and I have printed out all the feature map sizes. The problem I think is miss of resizing the input image. I also don't know the solution, and I change the 'valid' to 'train' and 'test' and find that the mAp is very low(13% for train and 0.6% for test). Is there any solution to the problem we have met?

HIi, I have met the same problem as yours, now, i run eval.py, map = 0.05, and some classes map is 0.
have you solved this problem? thanks.

from efficientdet.pytorch.

I also meet the same question, in validating , i changed the valid to 'test', and so, i get the mAp is 0.197 on VOC2012 Val.
The model was trained 187 epoches on Voc0712
last cls_loss : 0.1580013930797577
last reg_loss : 0.22156712412834167
last mean_loss : 0.5689369416236878
Can anybody help me ？ thanks

from efficientdet.pytorch.

I meet the same question, and I have printed out all the feature map sizes. The problem I think is miss of resizing the input image. I also don't know the solution, and I change the 'valid' to 'train' and 'test' and find that the mAp is very low(13% for train and 0.6% for test). Is there any solution to the problem we have met?

HIi, I have met the same problem as yours, now, i run eval.py, map = 0.05, and some classes map is 0.
have you solved this problem? thanks.

hi, have you solved the question? thanks

from efficientdet.pytorch.