RMPE: Regional Multi-person Pose Estimation
By Haoshu Fang, Shuqin Xie, Yu-Wing Tai, Cewu Lu. 2016
1,772 citations recorded by the source corpus.
Multi-person pose estimation in the wild is challenging. Although state-of-the-art human detectors have demonstrated good performance, small errors in localization and recognition are inevitable. These errors can cause failures for a single-person pose estimator (SPPE), especially for methods that solely depend on human detection results. In this paper, we propose a novel regional multi-person pose estimation (RMPE) framework to facilitate pose estimation in the presence of inaccurate human bounding boxes. Our framework consists of three components: Symmetric Spatial Transformer Network (SSTN), Parametric Pose Non-Maximum-Suppression (NMS), and Pose-Guided Proposals Generator (PGPG). Our method is able to handle inaccurate bounding boxes and redundant detections, allowing it to achieve 76:7 mAP on the MPII (multi person) dataset[3]. Our model and source codes are made publicly available.
Subfields: Computer Vision and Pattern Recognition, 3D Vision & Pose, Human-Computer Interaction
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Xiaogang Wang
- Yi Yang
- Trevor Darrell
- Jitendra Malik
- Andrew Zisserman
In 3D Vision & Pose
In Human-Computer Interaction
Citations in the corpus
- Objects as Points (2019, background): “Object detection powers many vision tasks like instance segmentation [7, 21, 32], pose estimation [3, 15, 39], tracking [24,27], and action recognition [5].”
- STAR: A Benchmark for Situated Reasoning in Real-World Videos (2024, methodology): “A pose parser (AlphaPose [9]) to extract skeleton representations for motions.”
- Deep High-Resolution Representation Learning for Visual Recognition (2019, neutral): “5 RMPE [38] PyraNet [165] 320× 256 28.”
- ResNeSt: Split-Attention Networks (2020, background): “ained for image classication often serve as the backbone of the neural networks designed for other applications, such as object detection [22,46], semantic segmentation [6,43,73] and pose estimation [14,58]. Recent work has signi- cantly boosted image classication accuracy through large scale neural”
- InstaBoost: Boosting Instance Segmentation via Probability Map Guided Copy-Pasting (2019, background): “COCO dataset is the state-of-the-art evaluation benchmark for computer vision tasks including bounding box detection [37], instance segmentation [28], human pose estimation [19] and captioning [39].”
- Monocular 3D Human Pose Estimation by Predicting Depth on Joints (2017, methodology): “In all model variations, we apply the pre-trained off-the-shelf 2D pose estimator from [9] to detect 2D poses without any re-training or fine-tuning because it is easy for the model to overfit the Human3.”
- Exploring Cross-Image Pixel Contrast for Semantic Segmentation (2021, neutral): “Our method generates promising results and shows great potential in a variety of dense prediction tasks, such as pose estimation [89, 21] and body parsing [88, 20].”
- HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation (2019, methodology): “HigherHRNet also outperforms the previous best method [24] (which performs a global refinement of top-down method [15]) by a healthy margin of 1.6 AP and most of the gain comes from AP M (+1.8 AP) and AP H (+1.5 AP), which contains images with the most crowd.”
- HAKE: Human Activity Knowledge Engine (2019, methodology): “We then use pose estimation [15] to obtain the pose keypoints of all the annotated persons, and their part bounding boxes following [14].”
- PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model (2018, background): “Examples of the former include G-RMI [33], CFN [35], RMPE [36], Mask R-CNN [34], and CPN [37].”
- All One Needs to Know about Metaverse: A Complete Survey on Technological Singularity, Virtual Ecosystem, and Research Agenda (2021, background): “human pose information from the RGB camera data [327]– [329] or 3D human pose information from RGB-D sensor data [330]–[332].”
- Pose Flow: Efficient Online Pose Tracking (2018, methodology): “Mainstream research fields have advanced from pose estimation of single pre-located person [6, 15] to multi-person pose estimation in complex and unconstrained scenes [3, 7].”
- Distribution-Aware Coordinate Representation for Human Pose Estimation (2019, methodology): “5 RMPE [11] PyraNet 320× 256 28.”
- Pose-Invariant Embedding for Deep Person Re-Identification (2017, methodology): “Further, when we use AlphaPose [46] for pose estimation and expand the length and width by 20%, rank-1 accuracy remains more or less the same: from 88.”
- Gaze360: Physically Unconstrained Gaze Estimation in the Wild (2019, methodology): “To build the dataset, we use AlphaPose [3] to detect the position of head keypoints and feet of subjects in rectified frames from each camera unit independently.”
- DiffusionDet: Diffusion Model for Object Detection (2022, background): “As a fundamental visual recognition task, it has become the cornerstone of many related recognition scenarios, such as instance segmentation [36, 53], pose estimation [9, 22], action recognition [32, 81], object tracking [46,65], and visual relationship detection [45,62].”
- Cascaded Pyramid Network for Multi-person Pose Estimation (2017, methodology): “Top-down approaches [28, 18, 15, 9] interpret the process of detecting keypoints as a twostage pipeline, that is, firstly locate and crop all persons from image, and then solve the single person pose estimation problem in the cropped person patches.”