-
自带的配置文件,比如YOLOv3_mobilenetv3_large_voc.yml和YOLOv3_mobilenetv3_ssld_large_voc.yml.除了预训练权重不一样,其他的参数都是一样的呀,难道知识蒸馏的实现就是通过预训练权重吗?
zmdcs updated
2 years ago
-
我在pytorch->onnx->ncnn对mobileNetv3模型转换的过程中,遇到了:1、onnx2ncnn步骤有不支持问题,使用了onnxsim对onnx模型进行了处理后,再进行onnx2ncnn时转换OK,得到解决;2、我在构建应用,使用extractor.extract()时需要blob的name或index,当我使用ncnn2mem将文件转换成*.id.h打开查看最后的index并使…
-
Activation _hard_swish in MobileNetV3_Small.
It's a segment model. Model is here!
[weights_01-0.8977.zip](https://github.com/MarsTechHAN/keras2ncnn/files/6269259/weights_01-0.8977.zip)
-
请问 fp16 的模型如何使用, fp32 的 cpu 下速度还是很慢
-
The number of the last block channel of mobilenetv3 in the paper is reduced by a factor of 2, and atrous conv is used
-
Hi, how can I add the biasloss loss function to mobilenetv3?
-
The [paper](https://arxiv.org/pdf/1905.02244.pdf) has been updated. The last but one 'bneck' is updated in 'Table 1'. The 'exp size' is no longer 672 but 960. Please check it out.
Is it possible to p…
-
thanks very much for your sharing, I got a bad results of training [mobilefacenet](https://github.com/deepinsight/insightface) with your mobilenetv3, how to improve? Part of the log content is below:…
-
问题类型:模型部署
**问题描述**
========================
利用官方提供的`yolov3_mobilenetv3_large_pruned`模型进行推理报错
```
/usr/lib/python3/dist-packages/apport/report.py:13: DeprecationWarning: the imp module is …
-
请问替换主干网络为mobilenetv3后,进行训练时,F1始终停留在0.49左右是什么原因?谢谢🙏
Sent from PPHub