movenet.pytorch icon indicating copy to clipboard operation
movenet.pytorch copied to clipboard

A Pytorch implementation of MoveNet from Google. Include training code and pre-trained model.

Results 36 movenet.pytorch issues
Sort by recently updated
recently updated
newest added

請問各位有沒有遇過在單色背景(如紅色)下,predict點位應該位於人物身上的被拉過去背景的情形?? 要如何解決?? 感謝 I met a situation like single color background such as red color, predict keypoints are located at red color not on human. Is there any solution for...

您好,您有对比过MoveNet和HRNet系列在coco上的精度对比吗?想做个基于骨骼关键点的步态识别,需要轻量化准确度比较好的姿态估计器,moveNet相比大型HRNet有差很多吗?

hey,i wanna you great pth like e100_valacc0.98349.pth,please help me。

在尝试使用别的数据集时,如Human3.6m这类若未标注是否被遮挡的数据集进行训练,应该如果处理呀? 如全标注为可见对结果的影响大吗?

FileNotFoundError: [Errno 2] No such file or directory: ../data/eval/mype . json' 大佬,是不是我处理数据的时候导致的问题,这个json文件要怎么生成

请问下regs译码的时候,为何要给regs_origin+0.5再int32呢,我验证了一张图片,其编码时对于48*48而言,最后一个关键点的标签是H W =42 16,但译码以后变成了42 17,原因就在于cx cy是24 24,原regsx本来是-8,但+0.5再int32导致其出现偏移,即对于x

Hi, currently i wanted to implement this into a real time using webcam. May i know is it possible? and how?

Hii, i want to train movenet with more than 17 keypoints. so what changes i have to do for training.

非常感谢您的工作,作为一个菜鸟想请问一下网络结构总体是mobilenetv2的backbone+fpn,但Mobilenetv2以matlab的结构版本或者最常见的版本是224的input,而且依据原论文中stride为1(可残差连接)的有1+3+1=5个,而在您的代码Mobilenetv2的forward中我注意到仅有3次残差连接,且原论文中mobilenetv2无上采样层unsample,这是什么原因导致您选择了将mobilnetv2进行了以下改动?或者是我理解错误?或者是说采用了Inverted Residuals这种mobilenet的特色结构的在业界均可称作以mobilenet为backbone