![](https://img-blog.csdnimg.cn/20201014180756925.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
pytorch深度学习
EericDL
deep learning work
展开
-
coco-eval-free and training with custom category
# coco-eval-free and training with custom categorysupporting Nanodet projects, https://github.com/RangiLyu/nanodet/blob/main/nanodet/evaluator/coco_detection.pycoco eval by custom category to multi object classificationif category is two, then id ind.原创 2020-12-09 10:58:26 · 139 阅读 · 1 评论 -
Nanodet project label: yolo2voc2coco
# yolo2voc2cocolabels: txt-->xml-->jsonThis project support to Nanodet project to make official labels.1.Nanodet project addresshttps://github.com/RangiLyu/nanodet2.making labels means2.1 alfred tool addresshttps://github.com/jinfagang/a原创 2020-12-01 11:43:12 · 324 阅读 · 0 评论 -
pytorch实现带标签格式数据的模型训练
使用带标签数据进行模型训练,训练架构是pytorch,包括数据读入,模型设计,训练过程,代码。原创 2017-10-20 09:56:11 · 4698 阅读 · 1 评论 -
pytorch基于卷积层通道剪枝的方法
本文基于文章:Pruning Convolutional Neural Networks for Resource Efficient Inferencehttps://arxiv.org/abs/1611.06440下面以ResNet模型为例,讲解剪枝方法。1 评估神经元(即通道)重要性1.1 给模型中所有conv layer重新编号排次序;1.2 计算每个conv ...原创 2017-10-21 09:16:50 · 7827 阅读 · 24 评论 -
pytorch实现LBCNN:Local Binary Convolutional Neural Networks
本文用pytorch实现,是基于发表在CVPR 2017上的文章:Local Binary Convolutional Neural Networks1.原文地址:https://arxiv.org/abs/1608.06049torch版开源代码:https://github.com/juefeix/lbcnn.torch2.详细介绍这篇文章内容的地址:http://...原创 2017-10-22 09:19:48 · 3189 阅读 · 5 评论