{"full_name":"PaddlePaddle/PaddleClas","owner":"PaddlePaddle","name":"PaddleClas","shard":"1","url":"https://github.com/PaddlePaddle/PaddleClas","description":"A treasure chest for visual classification and recognition powered by PaddlePaddle","homepage":"","topics":["image-classification","knowledge-distillation","autoaugment","randaugment","gridmask","cutmix","deit","repvgg","swin-transformer","pretrained-models","image-retrieval-system","image-recognition","product-recognition","fastdeploy"],"tag_pages":["image-classification","image-recognition","pretrained-models","swin-transformer"],"language":"Python","license":"Apache-2.0","avatar":"https://avatars.githubusercontent.com/u/23534030?v=4","created_at":"2020-03-30T04:27:42Z","pushed_at":"2026-11-12T06:19:36Z","metrics":{"stars":5863,"forks":1191,"watchers":76,"issues":1138,"prs":2300,"releases":11,"issues_open":111,"issues_closed":1027,"prs_open":11,"prs_closed":384,"prs_merged":1905,"discussions":3},"growth":{"day":{"stars":0,"forks":0,"base_day":"2026-11-15"}},"history":{"dates":["2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[5859,5861,5861,5861,5863,5863,5863],"forks":[1192,1192,1192,1192,1191,1191,1191],"watchers":[null,76,76,76,76,76,76],"issues_open":[null,111,111,111,111,111,111],"prs_open":[null,11,11,11,11,11,11]},"boards":{},"same_owner":[{"full_name":"PaddlePaddle/PaddleOCR","name":"PaddleOCR","shard":"1","boards":{"total":170}},{"full_name":"PaddlePaddle/Paddle-Lite","name":"Paddle-Lite","shard":"1","boards":{}},{"full_name":"PaddlePaddle/PaddleSeg","name":"PaddleSeg","shard":"1","boards":{}},{"full_name":"PaddlePaddle/PaddleSpeech","name":"PaddleSpeech","shard":"1","boards":{}}],"issues":[{"number":1535,"title":"为什么我使用ImageNet1k的train进行训练，在val上只能得到60.1%的ACC,按照论文和所给的预训练模型应该可以到66.8%的","comments":17,"created_at":"2021-12-07T11:33:04Z"},{"number":3338,"title":"官方样例demo报错: MemoryError: std::bad_alloc","comments":11,"created_at":"2025-01-03T10:30:46Z"},{"number":1162,"title":"使用release/2.0训练ResNet可以正常收敛，使用release/2.2不能收敛，均使用默认代码，只修改数据集位置","comments":11,"created_at":"2021-08-21T12:58:33Z"},{"number":1582,"title":"如何使用GPU进行数据增强处理","comments":10,"created_at":"2021-12-20T03:15:43Z"},{"number":2596,"title":"armeabi-v7a安卓无法编译","comments":9,"created_at":"2023-01-13T06:51:35Z"},{"number":2434,"title":"PP-ShiTuV2评估脚本","comments":8,"created_at":"2022-10-31T10:52:45Z"},{"number":1478,"title":"softmax的结果 sum 不等于1","comments":7,"created_at":"2021-11-24T10:55:56Z"},{"number":1202,"title":"个体识别训练问题","comments":7,"created_at":"2021-09-03T03:01:19Z"}],"discussions":[{"number":3181,"title":"PaddleClas正确率","comments":0,"category":"Q&A","created_at":"2024-07-08T03:50:19Z"},{"number":2809,"title":"请问 PP-HGNet 分类每个图片是只能有一个标签，还是可以有多个标签，如果可以有多个标签训练的txt文件格式怎么写？","comments":0,"category":"Q&A","created_at":"2023-05-21T14:29:36Z"},{"number":2346,"title":"language_classification","comments":0,"category":"General","created_at":"2022-09-25T19:03:26Z"}],"latest_release":{"name":"PaddleClas v2.6.0","tag":"v2.6.0","published_at":"2024-11-05T11:16:46Z","body":"- **🔥2024.11.5 添加图像分类和图像检索领域低代码全流程开发能力**:\r\n  *  飞桨低代码开发工具PaddleX，依托于PaddleClas的先进技术，支持了图像分类和图像检索领域的**低代码全流程**开发能力：\r\n     * 🎨 [**模型丰富一键调用**](docs/zh_CN/paddlex/quick_start.md)：将通用图像分类、图像多标签分类、通用图像识别、人脸识别涉及的**98个模型**整合为6条模型产线，通过极简的**Python API一键调用**，快速体验模型效果。此外，同一套API，也支持目标检测、图像分割、文本图像智能分析、通用OCR、时序预测等共计**200+模型**，形成20+单功能模块，方便开发者进行**模型组合使用**。\r\n     * 🚀 [**提高效率降低门槛**](docs/zh_CN/paddlex/overview.md)：提供基于**统一命令**和**图形界面**两种方式，实现模型简洁高效的使用、组合与定制。支持**高性能推理、服务化部署和端侧部署**等多种部署方式。此外，对于各种主流硬件如**英伟达GPU、昆仑芯、昇腾、寒武纪和海光**等，进行模型开发时，都可以**无缝切换**。\r\n  * 新增图像分类算法[**MobileNetV4、StarNet、FasterNet**](https://github.com/PaddlePaddle/PaddleX/blob/release/3.0-beta1/docs/module_usage/tutorials/cv_modules/image_classification.md)\r\n  * 新增服务端图像识别模型（图像特征）[**PP-ShiTuV2_rec_CLIP_vit_base、PP-ShiTuV2_rec_CLIP_vit_large**](https://github.com/PaddlePaddle/PaddleX/blob/release/3.0-beta1/docs/module_usage/tutorials/cv_modules/image_feature.md)\r\n  * 新增多标签图像分类模型[**CLIP_vit_base_patch16_448_ML、PP-HGNetV2-B0_ML、PP-HGNetV2-B4_ML、PP-HGNetV2-B6_ML、PP-LCNet_x1_0_ML、ResNet50_ML**](https://github.com/PaddlePaddle/PaddleX/blob/release/3.0-beta1/docs/module_usage/tutorials/cv_modules/ml_classification.md)\r\n  * 新增人脸识别模型**MobileFaceNet、ResNet50_face**，新增**人脸识别端到端系统**。","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}