PaddleClas is an image recognition and classification library built on the PaddlePaddle deep learning framework. It provides a collection of vision models including Swin-Transformer, DeiT, RepVGG, and MobileNetV4 for classification tasks, along with pretrained models and complete training workflows. The toolkit also supports knowledge distillation, data augmentation techniques like AutoAugment and CutMix, and deployment through FastDeploy for various hardware platforms.