PifPaf: Composite Fields for Human Pose Estimation

We propose a new bottom-up method for multi-person 2D human pose estimation that is particularly well suited for urban mobility such as self-driving cars and delivery robots. The new method, PifPaf, uses a Part Intensity Field (PIF) to localize body parts and a Part Association Field (PAF) to associate body parts with each other to form full human poses. Our method outperforms previous methods at low resolution and in crowded, cluttered and occluded scenes thanks to (i) our new composite field PAF encoding fine-grained information and (ii) the choice of Laplace loss for regressions which incorporates a notion of uncertainty. Our architecture is based on a fully convolutional, single-shot, box-free design. We perform on par with the existing state-of-the-art bottom-up method on the standard COCO keypoint task and produce state-of-the-art results on a modified COCO keypoint task for the transportation domain.

• Publication year

  2019

• Venue

  Computer Vision and Pattern Recognition

• Publication date

  2019-03-15

• Fields of study

  Computer Science, Engineering

• Identifiers
  DOI 10.1109/CVPR.2019.01225arXiv 1903.06593
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Multi-Person Pose Estimation via Multi-Layer Fractal Network and Joints Kinship Pattern

  2019cited by this paper
• Magnify-Net for Multi-Person 2D Pose Estimation

  2018cited by this paper
• MultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual Network

  2018cited by this paper
• DensePose: Dense Human Pose Estimation in the Wild

  2018cited by this paper
• Learning to Refine Human Pose Estimation

  2018cited by this paper
• Simple Baselines for Human Pose Estimation and Tracking

  2018cited by this paper
• PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model

  2018influential reference
• DeepLabCut: markerless pose estimation of user-defined body parts with deep learning

  2018cited by this paper
• What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?

  2017cited by this paper
• Cascaded Pyramid Network for Multi-person Pose Estimation

  2017cited by this paper
• Towards Accurate Multi-person Pose Estimation in the Wild

  2017cited by this paper
• Spindle Net: Person Re-identification with Human Body Region Guided Feature Decomposition and Fusion

  2017cited by this paper
• A Coarse-Fine Network for Keypoint Localization

  2017cited by this paper
• Mask R-CNN

  2017influential reference
• A Simple Yet Effective Baseline for 3d Human Pose Estimation

  2017cited by this paper
• Stacked Hourglass Networks for Human Pose Estimation

  2016cited by this paper
• Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network

  2016cited by this paper
• Convolutional Pose Machines

  2016cited by this paper
• Human Pose Estimation Using Deep Consensus Voting

  2016cited by this paper
• Person Re-identification by Multi-Channel Parts-Based CNN with Improved Triplet Loss Function

  2016cited by this paper
• Realtime Multi-person 2D Pose Estimation Using Part Affinity Fields

  2016influential reference
• Associative Embedding: End-to-End Learning for Joint Detection and Grouping

  2016cited by this paper
• DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

  2016cited by this paper
• Towards Viewpoint Invariant 3D Human Pose Estimation

  2016cited by this paper
• Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations

  2016cited by this paper
• Recurrent Human Pose Estimation

  2016cited by this paper
• RMPE: Regional Multi-person Pose Estimation

  2016cited by this paper
• DeeperCut: A Deeper, Stronger, and Faster Multi-person Pose Estimation Model

  2016cited by this paper
• DeepCut: Joint Subset Partition and Labeling for Multi Person Pose Estimation

  2015cited by this paper
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

  2015cited by this paper
• Human Pose Estimation with Iterative Error Feedback

  2015cited by this paper
• Deep Residual Learning for Image Recognition

  2015influential reference
• Flowing ConvNets for Human Pose Estimation in Videos

  2015cited by this paper
• Scalable Person Re-identification: A Benchmark

  2015influential reference
• Viewpoint Invariant Human Re-Identification in Camera Networks Using Pose Priors and Subject-Discriminative Features

  2015cited by this paper
• Fast R-CNN

  2015influential reference
• 2D Human Pose Estimation: New Benchmark and State of the Art Analysis

  2014cited by this paper
• Microsoft COCO: Common Objects in Context

  2014influential reference
• Human Pose Estimation Using Body Parts Dependent Joint Regressors

  2013cited by this paper
• DeepPose: Human Pose Estimation via Deep Neural Networks

  2013cited by this paper
• 2D Articulated Human Pose Estimation and Retrieval in (Almost) Unconstrained Still Images

  2012cited by this paper
• Object Detection with Discriminatively Trained Part Based Models

  2010cited by this paper
• Pictorial Structures for Object Recognition

  2004cited by this paper

• Self-Supervised 3D Human Localization using Canonical Pose

  2026cites this paper
• SD-HRNet: a lightweight high-resolution network for human pose estimation based on spatial decoupling

  2025cites this paper
• Dense symmetric temporal alignment learning for human pose estimation

  2025cites this paper
• Smart Yoga: Deep Learning-Based Framework for Yoga Pose Recognition and Real-Time Correction

  2025cites this paper
• Experiences Using MediaPipe to Make the Arms of a Humanoid Robot Imitate a Video-Recorded Dancer Performing a Robot Dance

  2025cites this paper
• WD2L: A novel approach for overcoming data imbalance in whole-body pose estimation

  2025cites this paper
• Group Inertial Poser: Multi-Person Pose and Global Translation from Sparse Inertial Sensors and Ultra-Wideband Ranging

  2025cites this paper
• Smart Gym Assistant: Posture Monitoring and Rep Counting

  2025cites this paper
• Comparing Human Pose Estimation through deep learning approaches: An overview

  2025cites this paper
• Human Pose Estimation from an Extremely Low-Resolution Image Sequence by Pose Transition Embedding Network

  2025cites this paper
• Generalizable large language model based human keypoint localization for emotion recognition

  2025cites this paper
• Instance-Specific Learning for Skeleton-Based Action Recognition with Varying Data Quality

  2025cites this paper
• Pedestrian Planet: What YouTube Driving from 233 Countries and Territories Teaches Us About the World

  2025cites this paper
• MythPose: Enhanced Detection of Complex Poses in Thangka Figures

  2025cites this paper
• Latent Diffusion-based Pedestrian Attention Estimation for Autonomous Driving

  2025influential citation
• Syncscore: A Framework for Synchronization Scoring in Group Sports Via Human Pose Estimation

  2025cites this paper
• Individual Action Detection using Yolov9 and Deep Neural Network

  2025cites this paper
• DualPose: Dual-Block Transformer Decoder with Contrastive Denoising for Multi-Person Pose Estimation

  2025cites this paper
• Human Pose Estimation: Single-Person and Multi-Person Approaches

  2025cites this paper
• MAQT: multi-scale attention and query-optimized transformer for end-to-end pose estimation

  2025cites this paper
• Multi-Person Pose Estimation with Feature Enhancement and Decoupling Based on Contrastive Learning

  2025cites this paper
• AI Driven Fall Detection: A Pose Estimation Approach

  2025cites this paper
• EPIC: Ego-Centric Monocular 3D Pedestrian Trajectory Estimation via Instance-Aware Center-Focused Depth Processing

  2025cites this paper
• Continuous Normalizing Flows for Uncertainty-Aware Human Pose Estimation

  2025cites this paper
• GaitMirror: Design and Implementation of an Artificial Intelligence-Based Gait Recognition System for Evaluating the Effectiveness of Lower Limb Rehabilitation

  2025cites this paper
• End-to-End Multi-Person Pose Estimation with Pose-Aware Video Transformer

  2025cites this paper
• DSP: Dense-Sparse Parallel Networks for Self-supervised 3D Multi-person Pose Estimation from Multiple Views

  2025cites this paper
• High-Resolution Spatiotemporal Modeling with Global-Local State Space Models for Video-Based Human Pose Estimation

  2025cites this paper
• Low-resolution human pose estimation and action recognition via pose-driven super-resolution reconstruction

  2025cites this paper
• Occlusion-Guided Feature Purification Learning via Reinforced Knowledge Distillation for Occluded Person Re-Identification

  2025cites this paper
• Graph Edge Classification for Keypoint Grouping in Multi-person Pose Estimation

  2025influential citation
• Assessing Traffic Conflicts Severity Through Simulated Collision Dynamics and Impact Analysis

  2025cites this paper
• MirrorPose: Enabling Full-Body Gestures Interaction for Head-Mounted Devices with a Full-Length Mirror

  2025cites this paper
• Semi-supervised Learning for Detector-free Multi-person Pose Estimation

  2025cites this paper
• BWFNet: 3D building reconstruction via building wireframe from single off-nadir image with semi-weak supervision

  2025cites this paper
• A CNN-Based Real-Time Pose Detection System Using MediaPipe for Human Activity Monitoring

  2025cites this paper
• Multi-Scale Position-Aware Attention and Irregular Erasure Network for Occluded Person Re-Identification

  2025cites this paper
• BR-Pose: enhancing human pose estimation through Bi-level routing attention and multi-level weight fusion

  2025cites this paper
• Shift Equivariant Pose Network

  2025cites this paper
• A hybrid human fall detection method based on modified YOLOv8s and AlphaPose

  2025cites this paper
• Adaptive Occlusion-Aware Network for Occluded Person Re-Identification

  2025cites this paper
• Improved YoloV5-Based Model for Instantaneous Smoke and Phone Use Detection during Dangerous Driving

  2025cites this paper
• SlimPose: Lightweight Multi-Person Pose Estimation via Multi-Scale Structural Feature Fusion and Selective Attention

  2025cites this paper
• Occlusion-aware and jitter-rejection 3D video real-time pose estimation for construction workers

  2025cites this paper
• Predicting Pedestrian Crossing Intention in Autonomous Vehicles: A Review

  2025cites this paper
• A Comprehensive Review of Real-Time Multi-View Multi-Person Markerless Motion Capture

  2025influential citation
• KeyRe-ID: Keypoint-Guided Person Re-Identification using Part-Aware Representation in Videos

  2025cites this paper
• Imitating human behavior to annotate key points for yak

  2025cites this paper
• VisualCent: Visual Human Analysis using Dynamic Centroid Representation

  2025cites this paper
• ROM-Pose: restoring occluded mask image for 2D human pose estimation

  2025cites this paper
• Keypoints as Dynamic Centroids for Unified Human Pose and Segmentation

  2025cites this paper
• PoseAlign network for hybrid structure in 2D human pose estimation

  2025cites this paper
• Few-shot learning-based human pose estimation model

  2025cites this paper
• SFG-YOLOv8: efficient and lightweight small-feature gesture keypoint detector

  2025cites this paper
• From Sharp to Blur: Unsupervised Domain Adaptation for 2D Human Pose Estimation Under Extreme Motion Blur Using Event Cameras

  2025cites this paper
• High-Resolution Human Keypoint Detection: A Unified Framework for Single and Multi-Person Settings

  2025cites this paper
• Social-Pose: Enhancing Trajectory Prediction With Human Body Pose

  2025cites this paper
• ST-TGR: Spatio-Temporal Representation Learning for Skeleton-Based Teaching Gesture Recognition

  2024cites this paper
• DHRNet: A Dual-Path Hierarchical Relation Network for Multi-Person Pose Estimation

  2024cites this paper
• GHNeRF: Learning Generalizable Human Features with Efficient Neural Radiance Fields

  2024cites this paper
• Video-Based Human Pose Regression via Decoupled Space-Time Aggregation

  2024cites this paper
• Enhancing Skeleton-Based Action Recognition in Real-World Scenarios Through Realistic Data Augmentation

  2024cites this paper
• ADVANCING YOGA PRACTICE : LEVERAGING OPENCV AND MEDIAPIPE FOR POSE DETECTION

  2024cites this paper
• SelfPose3d: Self-Supervised Multi-Person Multi-View 3d Pose Estimation

  2024cites this paper
• HF-HRNet: A Simple Hardware Friendly High-Resolution Network

  2024cites this paper
• Enhancing action recognition from low-quality skeleton data via part-level knowledge distillation

  2024cites this paper
• Real-Time Through-Wall Multiperson 3-D Pose Estimation Based on MIMO Radar

  2024cites this paper
• Multiple-Hand 2D Pose Estimation From a Monocular RGB Image

  2024cites this paper
• DROP: Decouple Re-Identification and Human Parsing with Task-specific Features for Occluded Person Re-identification

  2024cites this paper
• BP-YOLO: A Real-Time Product Detection and Shopping Behaviors Recognition Model for Intelligent Unmanned Vending Machine

  2024cites this paper
• GITPose: going shallow and deeper using vision transformers for human pose estimation

  2024cites this paper
• DRSI-Net: Dual-Residual Spatial Interaction Network for Multi-Person Pose Estimation

  2024cites this paper
• Real Time Fitness Tracking and Analysis using BlazePose Pose Estimation Ad

  2024cites this paper
• Enhancing Human Pose Estimation Accuracy Using Synthetic Data

  2024cites this paper
• Yoga Pose Detection Using Mediapipe and Cue Method

  2024cites this paper
• A Real-Time Yoga Instruction Method Based on Lightweight Human Pose Estimation

  2024cites this paper
• EFCPose: End-to-End Multi-Person Pose Estimation With Fully Convolutional Heads

  2024cites this paper
• Continuous Heatmap Regression for Pose Estimation via Implicit Neural Representation

  2024cites this paper
• UAGE: A Supervised Contrastive Method for Unconstrained Adaptive Gaze Estimation

  2024cites this paper
• SA-BiGCN: Bi-Stream Graph Convolution Networks With Spatial Attentions for the Eye Contact Detection in the Wild

  2024influential citation
• An integrated multi-person pose estimation and activity recognition technique using 3D dual network

  2024cites this paper
• Research Advanced in Human Pose Estimation based on Deep Learning

  2024cites this paper
• Distributed Dual-Quaternion Extended Kalman Filtering for Spacecraft Pose Estimation

  2024cites this paper
• OCDet: Object Center Detection via Bounding Box-Aware Heatmap Prediction on Edge Devices with NPUs

  2024cites this paper
• Research on an Improved RTMPose Model for Evaluating Dance Standardization Scores

  2024cites this paper
• Categorical Keypoint Positional Embedding for Robust Animal Re-Identification

  2024cites this paper
• Referring Human Pose and Mask Estimation in the Wild

  2024cites this paper
• Invisibility Cloak: Disappearance under Human Pose Estimation via Backdoor Attacks

  2024cites this paper
• HFGaussian: Learning Generalizable Gaussian Human with Integrated Human Features

  2024cites this paper
• Pedestrian Crossing Intention Prediction in the Wild: A Survey

  2024cites this paper
• CFAF-YOLOv8pose: Improved YOLOv8pose with Channel Feature Aggregation and Asymptotic Feature Fusion for Multi-person Human Pose Estimation

  2024cites this paper
• Improved Real-time Yoga Pose Estimation with GAN Augmentation

  2024cites this paper
• Human Skeleton Point Detection Based on Super Resolution

  2024cites this paper
• Pruning network for human pose estimation based on multiscale cross-attention

  2024cites this paper
• Spatial-Aware Regression for Keypoint Localization

  2024cites this paper
• Computer Vision-Based Human Body Posture Correction System

  2024cites this paper
• PAFormer: Part Aware Transformer for Person Re-identification

  2024cites this paper
• Instance-Level Data Augmentation for Multi-Person Pose Estimation: Improving Recognition of Individuals at Different Scales

  2024cites this paper
• Video Interaction Recognition using an Attention Augmented Relational Network and Skeleton Data

  2024cites this paper
• Frontal person image generation based on arbitrary‐view human images

  2024cites this paper