{"id":13486541,"url":"https://github.com/lxztju/pytorch_classification","last_synced_at":"2025-05-16T06:07:30.192Z","repository":{"id":37675663,"uuid":"256963384","full_name":"lxztju/pytorch_classification","owner":"lxztju","description":"利用pytorch实现图像分类的一个完整的代码，训练，预测，TTA，模型融合，模型部署，cnn提取特征，svm或者随机森林等进行分类，模型蒸馏，一个完整的代码","archived":false,"fork":false,"pushed_at":"2023-02-06T13:30:18.000Z","size":3209,"stargazers_count":1408,"open_issues_count":41,"forks_count":339,"subscribers_count":13,"default_branch":"master","last_synced_at":"2025-04-08T16:05:22.419Z","etag":null,"topics":["cnn","densenet","deployment","flask","image-classification","knn","knowledge-distillation","label-smoothing","pytorch","random-forest","resnet","resnext","svm"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/lxztju.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2020-04-19T09:43:18.000Z","updated_at":"2025-04-07T13:26:35.000Z","dependencies_parsed_at":"2024-05-21T00:44:41.504Z","dependency_job_id":null,"html_url":"https://github.com/lxztju/pytorch_classification","commit_stats":null,"previous_names":[],"tags_count":2,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lxztju%2Fpytorch_classification","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lxztju%2Fpytorch_classification/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lxztju%2Fpytorch_classification/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lxztju%2Fpytorch_classification/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/lxztju","download_url":"https://codeload.github.com/lxztju/pytorch_classification/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":254478190,"owners_count":22077676,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["cnn","densenet","deployment","flask","image-classification","knn","knowledge-distillation","label-smoothing","pytorch","random-forest","resnet","resnext","svm"],"created_at":"2024-07-31T18:00:48.056Z","updated_at":"2025-05-16T06:07:25.176Z","avatar_url":"https://github.com/lxztju.png","language":"Jupyter Notebook","funding_links":[],"categories":["Jupyter Notebook"],"sub_categories":[],"readme":"\n## 简介\n\n基于torchision实现的pytorch图像分类功能。\n\n\n## 近期更新\n\n* 2022.11.05更新\n    - 新添加tensorrt c++的推理方案\n\n* 2022.10.29更新，进行代码重构，基本的功能基本一致。\n    - 支持pytorch ddp的训练\n    - 支持c++ libtorch的模型推理\n    - 支持script脚本一键运行\n    - 添加日志模块\n\n习惯之前版本的请看v1版本的代码：[V1版本](https://github.com/lxztju/pytorch_classification/tree/v1)。\n\n\n主要功能：\n\n利用pytorch实现图像分类，基于torchision可以扩展使用densenet，resnext，mobilenet，efficientnet，swin transformer等图像分类网络\n\n如果有用欢迎star\n\n## 实现功能\n* 基础功能利用pytorch实现图像分类\n* 包含带有warmup的cosine学习率调整\n* warmup的step学习率优调整\n* 多模型融合预测，加权与投票融合\n* 利用flask + redis实现模型云端api部署（tag v1）\n* c++ libtorch的模型部署\n* 使用tta测试时增强进行预测（tag v1）\n* 添加label smooth的pytorch实现（标签平滑）（tag v1）\n* 添加使用cnn提取特征，并使用SVM，RF，MLP，KNN等分类器进行分类（tag v1）。\n* 可视化特征层\n\n## 运行环境\n* python3.7\n* pytorch 1.8.1\n* torchvision 0.9.1\n* opencv(libtorch cpp推理使用， 版本3.4.6)（可选）\n* libtorch cpp推理使用（可选）\n\n\n\n## 快速开始\n\n### 数据集形式\n 数据集的组织形式，参考[sample_files/imgs/listfile.txt](https://github.com/lxztju/pytorch_classification/blob/master/sample_files/imgs/listfile.txt)\n\n\n### 训练 测试\n\n修改`run.sh`中的参数，直接运行run.sh即可运行\n\n\n主要修改的参数：\n\n```\nOUTPUT_PATH 模型保存和log文件的路径\n\nTRAIN_LIST 训练数据集的list文件\nVAL_LIST  测试集合的list文件\nmodel_name 默认是resnet50\nlr 学习率\nepochs 训练总的epoch\nbatch-size  batch的大小\nj dataloader的num_workers的大小\nnum_classes 类别数\n```\n\n\n### libtorch inference\n\n\n代码存储在`cpp_inference`文件夹中。\n\n1. 利用[cpp_inference/traced_model/trace_model.py](https://github.com/lxztju/pytorch_classification/blob/master/cpp_inference/traced_model/trace_model.py)将训练好的模型导出。\n2. 编译所需的opencv和libtorch代码到`cpp_inference/third_party_library`\n\n3. 编译\n```\nsh compile.sh\n```\n\n4. 可执行文件测试\n```\n./bin/imgCls imgpath\n```\n\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flxztju%2Fpytorch_classification","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Flxztju%2Fpytorch_classification","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flxztju%2Fpytorch_classification/lists"}