Pytorch_Retinaface
Pytorch_Retinaface copied to clipboard
biubug6
Retinaface get 80.99% in widerface hard val using mobilenet0.25.
Hi, the fddb eval tool mentioned in README is not accessible, chould you please update the link?
I use following cmd to do evaluation > python test_widerface.py --trained_model mobilenet0.25_Final.pth --network mobile0.25 but it turns out that "mobilenetV1X0.25_pretrain.tar" is used([here](https://github.com/biubug6/Pytorch_Retinaface/blob/b984b4b775b2c4dced95c1eadd195a5c7d32a60b/models/retinaface.py#L60)) and it is the imagenet pretrain weights, which...
import torch import torch.nn as nn import torchvision.models.detection.backbone_utils as backbone_utils import torchvision.models._utils as _utils import torch.nn.functional as F from collections import OrderedDict from layers.functions.prior_box import PriorBox from models.net import MobileNetV1...
Can you please your five facial landmark annotating tools? I am working on my project, where I am going to use the RetinaFace, but performance is not good enough while...
我在将retinaface 转onnx 转 mxnet 的过程中，发现 ONNX会将 F.interpolate (mode='nearest')转成 resize op，但是mxnet 又不支持resize op，不知道有没有同学遇到过这个问题？
loc_t = torch.Tensor(num, num_priors, 4) landm_t = torch.Tensor(num, num_priors, 10) conf_t = torch.LongTensor(num, num_priors) 我打印了一下，loc_t，landm_t ， conf_t初始值都不是0，如果后面赋值有漏洞，loss可能有问题。
I have been trying to use this repo to detect faces but it seems that it detects a lot of them where there aren't any in an image if used...
Thank you for your very useful tool. I wanted to use a retina face for my project related to face detection from the CCTV data after anonymizing them. But detection...
In my experience, retinaface cannot precisely locate the 5 landmarks on rotated faces, especially for 90 degrees and more rotation. How to solve that problem without rotating the input image?...
