EfficientDet.Pytorch

EfficientDet.Pytorch copied to clipboard

The demo.py has too many boxes to incorrect, and the code updated is unstable.

5

y78h11b09

Why are there two thresholds?

3

Can you say what these two thresholds represent? threshold and iou_threshold Thank you！！！！

ma3252788

changed on d3 and d4 for 3 days, result divergent

1

There were no box output when evaluation.

lucasjinreal

In the file models/efficientnets.py there is two from_pretrained() methods, is it normal ? Whats the point ?

Cyril9227

In this implementation as well as others I've found online, there are separate weights for each layer for the feature fusion: https://github.com/toandaominh1997/EfficientDet.Pytorch/blob/fbe56e58c9a2749520303d2d380427e5f01305ba/models/bifpn.py#L147 In [the paper](https://arxiv.org/pdf/1911.09070.pdf) it appears as though the...

oadams

Q1: https://github.com/toandaominh1997/EfficientDet.Pytorch/blob/master/models/efficientdet.py#L33-L53 ``` python self.backbone = EfficientNet.from_pretrained(MODEL_MAP[network]) ... for m in self.modules(): if isinstance(m, nn.Conv2d): n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels m.weight.data.normal_(0, math.sqrt(2. / n)) elif isinstance(m, nn.BatchNorm2d): m.weight.data.fill_(1)...

yichengsu

AttributeError: 'EfficientDet' object has no attribute 'module'

4

** Please help me. Model not found after start of training This problem occurs when I specify GPU。 Setting the GPU configuration to none will not cause this problem, but...

wangtianlong1994

Hi! Thank you for a great implementation. I noticed one strange thing: Albumentation's Resize doesn't scale a bbox. It's written (the source code documentation) that the bbox is scale invariant....

AldusLorde

Thank you very much for sharing. Does the data have a validation set? I did not see it.

sihuo521

Images fed to the network in demo.py are BGR instead of RGB. This is apparent in the image in your readme - the car's headlights are blue, as are the...

avantgardnerio