EfficientDet.Pytorch issues

Results 101 EfficientDet.Pytorch issues

Sort by recently updated

threshold/iou_threshold

Just wanted to ask if people knew the meaning behind each of these "threshold" flags -- why are there two? where do they each appear in the source code?

alexrider1105

The MAP of auther's pre-trained model EfficientDet-d0 on VOC is about 0.56?

I have run the script eval.py with the model EfficientDet-d0 offered by author, but the MAP is only about 0.56. Do you know why? I hope you can help me....

majian-stu

efficientnet-b0 weights url is broken

http://storage.googleapis.com/public-models/efficientnet/efficientnet-b0-355c32eb.pth this url doesn't work anymore... HTTP Error 403: Forbidden from where can I download valid weights?

ZaalDz

No Depth-wise convolution for BiFPN?

The EfficientDet paper uses depth-wise convolution of BiFPN, however, in this implementation, clearly depth-wise convolution is not used (in module.py). `conv_cfg = { 'Conv': nn.Conv2d, 'ConvWS': ConvWS2d, # TODO: octave...

yaoliUoA

Improve class number calculation for COCO

Substitute the previous hard-coded value `80` with a more robust version.

lolipopshock

To test the VOC dataset, the mAP is too low, and many categories mAP are 0

Dear, beloved, the world's best author, excuse me, can you do VOC well first?When I use demo.py, the results are also very poor.Please, the best engineer in the universe

944284742

update URL mapping and add advprop pretrain models

@toandaominh1997 thank's for your amazing work. However the URL links are broken, this PR fixes the URL mapping and also provides the option to load pretrained models using adversarial training,...

Cyril9227

No boxes to NMS

Hi, I have trained using : python3 train.py --dataset COCO --dataset_root data/coco/ --network efficientdet-d0 --batch_size 8 --num_epoch 15 --num_class 1 --save_folder saved/ After 5 epochs: > > cls_loss : 2.9468758384609828e-06...

SaiAinala3