weilanShi

Results 35 comments of weilanShi

Hello, I have succeeded in converting. weigths to. tflite. The steps are as follows: first, convert. weights. cfg to. pb and .ckpt, referring specifically to: https://github.com/jinyu121/DW2TF/issues/30 https://github.com/jinyu121/DW2TF/releases/tag/freez/tensorflow tool e_graph, then...

不好意思,后面没有选用这个,没测试过 ---原始邮件--- 发件人: ***@***.***> 发送时间: 2022年3月7日(周一) 中午11:37 收件人: ***@***.***>; 抄送: ***@***.******@***.***>; 主题: Re: [zeusees/HyperFAS] 推理速度 (Issue #18) 你好,请问您做了这个项目了么 可以交流一下么 — Reply to this email directly, view it on GitHub, or...

> 你的pb我看了一下,没有不支持的op,你这个h5 netron都打不开,一般来说肯定是转换出问题了。 > 我明天帮你转一下试试看吧,tf这个一堆模型转换的问题是真的头疼= = 感谢大佬!tf 想在 ncnn下跑真的是太难了(就是钟情于ncnn, 莫得办法)

> 1. fp16->int8 是二分之一 > 2. 使用最新版本的ncnn > 3. yolov5 使用 swish 激活,这无法量化导致频繁退回 fp16 计算,建议改用 relu 激活 你好,int8量化时 Prelu 会有问题吗

> 用自己的数据集训练出的模型,检测结果有时会出现置信度和类别为-1的情况,大部分情况下正常。 > 这是out输出结果:{'detection_out': array([[[[ 0., -1., -1., -1., -1., -1., -1.]]]], dtype=float32)} > box:[[-1280 -720 -1280 -720]](我原图是1280x720) > cls:[-1.] > conf:[- > 用自己的数据集训练出的模型,检测结果有时会出现置信度和类别为-1的情况,大部分情况下正常。 > 这是out输出结果:{'detection_out': array([[[[ 0., -1., -1.,...

> Hi, @chuanqi305 Thanks for your great job, I got some problems here. I used scrips to convert tf model to caffe model and got "deploy.caffemodel". After that, I used...

> i replaced these two file with the original files in caffe, now the issue was gone. i am not familiar with cuda, i am using Tesla M40 with 24GB...

> thanks for your great work! i am using your train.prototxt to train my dataset and finetunn with your deploy_vox.caffemodel, but loss decreases from 16 to 11 with several steps,...

the same problem..have you solved it?

> > > > > > > > @yuffon It's all the same with the original code, the only part I change is "x = conv_layer(x, filter=in_channel*0.5, kernel=[1,1], layer_name=scope+'_conv1')" >...