Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Xin Yan eba11d7266 | 1 year ago | |
---|---|---|
PaddleClas | 1 year ago | |
model | 1 year ago | |
.gitattributes | 1 year ago | |
LICENSE | 1 year ago | |
README.md | 1 year ago | |
test.py | 1 year ago | |
train.py | 1 year ago | |
train_demo.py | 1 year ago |
商品识别是零售结算的关键部分, 本仓库采用基于PP-ShiTu的"主体检测、特征提取、向量检索"构成的图像识别技术. 其中, 特征提取部分对模型的识别结果至关重要. 本仓库存储着ColugoMum项目团队长时间以来的实验成果, 基于RP2K数据集进行的一些重要实验, 可供开发者基于这些实验结论和模型进行模型的优化. 并且, 本仓库也提供针对不同场景的高精度推理模型, 开发者可直接下载推理模型进行业务开发.
零售场景对于精度和速度有着极高的要求, 其要求模型在确保精度的同时又要保证预测速度. 因此, 本仓库的模型选择会严格选取业内在速度与精度的平衡上有较好效果的模型. 相关模型的介绍可以参照本仓库百科, 详细介绍请参考其论文.
RP2K数据集收录了50万+张零售商品货架图片, 商品种类超过2, 000种, 是目前零售类数据集中产品种类数量最多的数据集. 不同于一般聚焦新产品的数据集, RP2K收录了超过50万张零售商品货架图片, 商品种类超过2000种, 该数据集是目前零售类数据集中产品种类数量TOP1, 同时所有图片均来自于真实场景下的人工采集, 针对每种商品, 品览提供了十分详细的标注. RP2K致力于帮助物品识别领域进行学术研究, 同时为AI物品识别从业者打造真实行业级试炼场.
在真实场景中, 准确识别货架上零售产品仍然具有很高的挑战性.
model | num epoch | batch size/gpu cards | embedding size | learning rate | cutout | RandomErasing | 测试分辨率 | top1 recall | 配置文件 |
---|---|---|---|---|---|---|---|---|---|
PP_LCNet_x2_5 | 100 | 256/1 | 512 | 0.02 | N | N | 224 | 96.40% | 配置文件 |
PP_LCNet_x2_5 | 100 | 256/1 | 512 | 0.02 | N | N | 288 | 96.76% | 配置文件 |
PP_LCNet_x2_5 | 100 | 256/1 | 512 | 0.02 | Y | N | 288 | 96.90% | 配置文件 |
PP_LCNet_x2_5 | 100 | 256/1 | 4096 | 0.02 | N | N | 288 | 96.79% | 配置文件 |
PP_LCNet_x2_5 | 100 | 256/1 | 4096 | 0.02 | N | Y | 288 | 96.87% | 配置文件 |
PP_LCNet_x2_5 | 100 | 256/1 | 4096 | 0.02 | Y | N | 288 | 96.93% | 配置文件 |
model | num epoch | batch size/gpu cards | embedding size | learning rate | top1 recall | 配置文件 | 预训练模型 | 推理模型 |
---|---|---|---|---|---|---|---|---|
PP_LCNet_x2_5 | 100 | 256/1 | 512 | 0.02 | 96.97% | 配置文件 | 预训练模型 | 推理模型 |
PP_LCNet_x2_5 | 100 | 256/1 | 4096 | 0.02 | 96.91% | 配置文件 | 预训练模型 | 推理模型 |
PP_LCNetV2_base | 120 | 256/1 | 512 | 0.02 | 96.92% | 配置文件 | 预训练模型 | 推理模型 |
PPHGNet_tiny | 120 | 256/1 | 4096 | 0.02 | 96.87% | 配置文件 | 预训练模型 | 推理模型 |
PPHGNet_small | 120 | 256/1 | 4096 | 0.02 | 96.87% | 配置文件 | 预训练模型 | 推理模型 |
PPHGNet_base | 120 | 256/1 | 4096 | 0.02 | 96.88% | 配置文件 | 预训练模型 | 推理模型 |
注:
欢迎开发者们加入本仓库的共建! ColugoMum渴望和开发者们一道, 推动我国AI+零售行业新变革!
@InProceedings{Le_2020_ECCV,
author = {Lele Cheng and Xiangzeng Zhou and Liming Zhao and Dangwei Li and Hong Shang and Yun Zheng and Pan Pan and Yinghui Xu.},
title = {Weakly Supervised Learning with Side Information for Noisy Labeled Images},
booktitle = {The European Conference on Computer Vision (ECCV)},
month = {August},
year = {2020}
}
@article{peng2020rp2k,
title={RP2K: A Large-Scale Retail Product Dataset forFine-Grained Image Classification},
author={Peng, Jingtian and Xiao, Chang and Li, Yifan},
journal={arXiv preprint arXiv:2006.12634},
year={2020}
}
@misc{cui2021pplcnet,
title={PP-LCNet: A Lightweight CPU Convolutional Neural Network},
author={Cheng Cui and Tingquan Gao and Shengyu Wei and Yuning Du and Ruoyu Guo and Shuilong Dong and Bin Lu and Ying Zhou and Xueying Lv and Qiwen Liu and Xiaoguang Hu and Dianhai Yu and Yanjun Ma},
year={2021},
eprint={2109.15099},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
@misc{yu2021pppicodet,
title={PP-PicoDet: A Better Real-Time Object Detector on Mobile Devices},
author={Guanghua Yu and Qinyao Chang and Wenyu Lv and Chang Xu and Cheng Cui and Wei Ji and Qingqing Dang and Kaipeng Deng and Guanzhong Wang and Yuning Du and Baohua Lai and Qiwen Liu and Xiaoguang Hu and Dianhai Yu and Yanjun Ma},
year={2021},
eprint={2111.00902},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
存放商品识别模型及训练日志,并对开发者开放,方便开发者基于此进行调参优化。
Text Markdown Python C++ Shell other
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》