PaddleClas v2.6.0

• 🔥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:
  • 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：
    • 🎨 模型丰富一键调用：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。
    • 🚀 提高效率降低门槛：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。
  • 新增图像分类算法MobileNetV4、StarNet、FasterNet
  • 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
  • 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
  • 新增人脸识别模型MobileFaceNet、ResNet50_face，新增人脸识别端到端系统。

PaddleClas v2.5.2

What's Changed

• [cherry-pick] rm the faiss version by @TingquanGao in #2554
• Update data_augmentation.md by @weisy11 in #2566
• [cherry pick]fix retrieval for windows by @HydrogenSulfate in #2581
• [cherry-pick] fix doc by @TingquanGao in #2592
• [cherry-pick] docs: fix by @TingquanGao in #2611
• Update README_ch.md by @Evezerest in #2619
• [cherry-pick] fix: set dtype in paddle.to_tensor() by @TingquanGao in #2643
• [cherry pick]Add android demo doc to quick start by @HydrogenSulfate in #2672
• [cherry-pick] fix acc metric & fix format by @TingquanGao in #2607
• [FastDeploy] Update README_ch.md by @yunyaoXYY in #2708
• [NPU] update npu inference api by @YanhuiDua in #2775
• [cherry-pick] Update QR code by @TingquanGao in #2833
• docs: rm specified branch by @TingquanGao in #2846
• [cherry-pick] adapt to Pillow new version 10.0.0 by @TingquanGao in #2859
• [npu-tipc] fix npu tipc scripts to reduce running time(#2851) by @YanhuiDua in #2852
• [NPU TIPC] update npu python version in tipc config(#2885) by @YanhuiDua in #2884
• change .numpy()[0] to float() for correct usage of 0-D tensor by @HydrogenSulfate in #2905
• update readme for paddlex by @cuicheng01 in #2953
• Update utils.py by @awarebayes in #2886
• [cherry-pick]cherry-pick tipc scripts update and fix np.int error by @USTCKAY in #3011
• update readme for paddlex by @cuicheng01 in #3028
• update paddlex of readme by @zhangyubo0722 in #3074
• add mirrors-yapf by @changdazhou in #3096
• Add a script for evaluating all models & fix errors in some configuration files by @Tongkaio in #3101
• try using github action to automate release package uploading by @jzhang533 in #3113

New Contributors

• @awarebayes made their first contribution in #2886
• @Tongkaio made their first contribution in #3101
• @jzhang533 made their first contribution in #3113

Full Changelog: v2.5.1...v2.5.2

PaddleClas v2.5.1

1. support build index by whl;
2. fix some bugs.

PaddleClas v2.5.0

1.Release PP-ShiTu V2.
2.Release PP-ShiTu V2 android demo.
3.Release PP-ShiTu feature database management tool.

PaddleClas v2.4.0

1.Release Practical Ultra Light-weight image Classification solutions. PULC models inference within 3ms on CPU devices, with accuracy on par with SwinTransformer.
2.Release 9 PULC models including person attribute, traffic sign recognition, text image orientation classification, etc.
3.Release PP-HGNet classification network, which is suitable for gpu devices
4.Release PP-LCNet v2 classification network, which is suitable for cpu devices.
5.Add CSwinTransformer, PVTv2, MobileViT and VAN.
6.Add BoT ReID models.

PaddleClas v2.3.1

1.Update PP-ShiTu model and add 18MB model series.
2.Upgrade the document completely.
3.Add C++ Inference.
4.Add C++ Pipeline Serving mode.
5.Add a demo for Paddle Lite on Android.

PaddleClas v2.3.0

1.Add lite weight models, including detection and feature extraction.
2.Add PP-LCNet backbone model, which is super fast on CPU devices.
3.Support PaddleServing and PaddleSlim.
4.Switch Vector Search module to faiss, due to many compatibility feedback.
5.Add PKSampler, which is more stable on multi-card training.
6.Legendary models now can output middleware result.
7.Add DeepHash module, which can compress float feature to binary feature.
8.SwinTransformer, Twins and Deit achieve same accuracy with the origins training from scratch.

PaddleClas v2.2.1

1.Add Swin transformer series model.
2.Support static graph training, support dali and fp16 training.
3.Support build feature gallery with batchsize > 1, support add new feature to existing feature gallery.
4.Fix bugs and update document.

PaddleClas v2.2.0

1. Architecture
   1.1. ppcls backbones are now separated into two groups: legendary models and model zoo.
   1.2. Legendary models inherit from a new base class TheseusLayer, which allows stop at some point or even change architectures
2. Metric Learning
   2.1. Add a lot of metric learning functions, including gears, which can be inserted into arch , and losses.
   2.2. PaddleClas now support classification task and metric learning task using the same trainer. You only need switch different configs.
3. Vector Search
   3.1. Intergrate Mobius vector search algorithm.
4. Applications
   4.1. Add new applications: product recognition, logo recognition, car classification, car ReID and cartoon character recognition.
   4.2. Add new image recognition pipeline, which contains detection, feature extraction and vector search.
5. New models
   5.1. add LeViT、Twins、TNT、DLA、HarDNet、RedNet models

PaddleClas v2.1.0

1. Add RexNet, Mixnet, ViT and DeiT deploy model.
2. Add new Chinese tutorials for different users.
3. Update whl package.