license: apache-2.0
pipeline_tag: keypoint-detection
library_name: pytorch
tags:
- model_hub_mixin
- pytorch_model_hub_mixin
- vision
- pose-estimation
EdgeCrafter (ECPose)
EdgeCrafter is a unified compact Vision Transformer (ViT) framework designed for edge dense prediction tasks like object detection, instance segmentation, and pose estimation. This specific checkpoint is for ECPose, a variant specialized for high-performance human pose estimation on resource-constrained devices.
- Paper: EdgeCrafter: Compact ViTs for Edge Dense Prediction via Task-Specialized Distillation
- Code: GitHub Repository
- Project Page: EdgeCrafter Project Page
Model Description
Deploying high-performance dense prediction models on edge devices is challenging due to strict limits on computation and memory. EdgeCrafter addresses this gap using task-specific representation learning and a distilled compact backbone.
For pose estimation, the ECPose-X variant reaches 74.8 AP on the COCO dataset, significantly outperforming YOLO-based architectures (e.g., YOLOv8-Pose-X at 71.6 AP) while maintaining an efficiency profile suitable for edge deployment.
Usage
This model has been pushed to the Hub using the PytorchModelHubMixin integration.
For inference examples and detailed setup, please refer to the official GitHub repository. A CLI inference example is provided below:
# Refer to GitHub for installation and config requirements
cd ecpose
python tools/inference/torch_inf.py -c configs/ecpose/ecpose_s_coco.yml -r ecpose_s.pth -i path/to/your/image.jpg
Citation
If you find EdgeCrafter useful in your research, please consider citing:
@article{liu2026edgecrafter,
title={EdgeCrafter: Compact ViTs for Edge Dense Prediction via Task-Specialized Distillation},
author={Liu, Longfei and Hou, Yongjie and Li, Yang and Wang, Qirui and Sha, Youyang and Yu, Yongjun and Wang, Yinzhi and Ru, Peizhe and Yu, Xuanlong and Shen, Xi},
journal={arXiv},
year={2026}
}