This paper presents simple and effective baseline methods for human pose estimation and tracking, achieving state-of-the-art results on challenging benchmarks. The methods are based on a deconvolutional network added to a ResNet backbone for pose estimation, and a flow-based tracking approach for pose tracking. The pose estimation method uses a few deconvolutional layers to generate heatmaps from deep and low-resolution feature maps, achieving a mAP of 73.7 on the COCO test-dev split, which is an improvement over previous methods. The pose tracking method uses optical flow-based pose propagation and similarity measurement, achieving a mAP of 74.6 and a MOTA score of 57.8, which is an improvement over the ICCV'17 PoseTrack Challenge winner. The methods are validated through comprehensive ablation studies and are shown to be effective and simple. The code and pre-trained models are available for research purposes. The paper also discusses the importance of high-resolution feature maps in pose estimation and the effectiveness of flow-based pose similarity in pose tracking. The methods are compared with state-of-the-art approaches on the COCO and PoseTrack datasets, showing their superiority in performance. The results demonstrate that simple methods can achieve state-of-the-art results, and the baselines provide a useful reference for future research.This paper presents simple and effective baseline methods for human pose estimation and tracking, achieving state-of-the-art results on challenging benchmarks. The methods are based on a deconvolutional network added to a ResNet backbone for pose estimation, and a flow-based tracking approach for pose tracking. The pose estimation method uses a few deconvolutional layers to generate heatmaps from deep and low-resolution feature maps, achieving a mAP of 73.7 on the COCO test-dev split, which is an improvement over previous methods. The pose tracking method uses optical flow-based pose propagation and similarity measurement, achieving a mAP of 74.6 and a MOTA score of 57.8, which is an improvement over the ICCV'17 PoseTrack Challenge winner. The methods are validated through comprehensive ablation studies and are shown to be effective and simple. The code and pre-trained models are available for research purposes. The paper also discusses the importance of high-resolution feature maps in pose estimation and the effectiveness of flow-based pose similarity in pose tracking. The methods are compared with state-of-the-art approaches on the COCO and PoseTrack datasets, showing their superiority in performance. The results demonstrate that simple methods can achieve state-of-the-art results, and the baselines provide a useful reference for future research.