This paper presents a strong baseline for person re-identification (ReID) using simple and effective training tricks. The proposed method achieves 94.5% rank-1 accuracy and 85.9% mAP on the Market1501 dataset using only global features. The baseline is based on a standard ResNet50 backbone with several training tricks applied, including warmup learning rate, random erasing augmentation, label smoothing, stride change, BNNeck, and center loss. These tricks are evaluated and combined to improve the performance of the ReID model. The BNNeck structure is designed to separate features into different subspaces, improving the performance of the model. The results show that the proposed baseline outperforms many state-of-the-art methods in terms of performance, especially when using only global features. The paper also discusses the importance of considering training tricks when evaluating academic papers and highlights the benefits of using simple and effective models in industry. The code and models are available for download.This paper presents a strong baseline for person re-identification (ReID) using simple and effective training tricks. The proposed method achieves 94.5% rank-1 accuracy and 85.9% mAP on the Market1501 dataset using only global features. The baseline is based on a standard ResNet50 backbone with several training tricks applied, including warmup learning rate, random erasing augmentation, label smoothing, stride change, BNNeck, and center loss. These tricks are evaluated and combined to improve the performance of the ReID model. The BNNeck structure is designed to separate features into different subspaces, improving the performance of the model. The results show that the proposed baseline outperforms many state-of-the-art methods in terms of performance, especially when using only global features. The paper also discusses the importance of considering training tricks when evaluating academic papers and highlights the benefits of using simple and effective models in industry. The code and models are available for download.