# Slim DensePose: Thrifty Learning from Sparse Annotations and Motion Cues

arXiv: https://arxiv.org/abs/1906.05706 · HTML: https://mlsanity.com/paper/1906.05706/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Jun 2019.

## Built on by (most cited first, 13 of 13)

- [Learning monocular 3D reconstruction of articulated categories from motion](https://mlsanity.com/paper/2103.16352/) (2021)
- [EvHandPose: Event-based 3D Hand Pose Estimation with Sparse Supervision](https://mlsanity.com/paper/2303.02862/) (2023)
- [Discovering Relationships between Object Categories via Universal Canonical Maps](https://mlsanity.com/paper/2106.09758/) (2021)
- [HumanGPS: Geodesic PreServing Feature for Dense Human Correspondences](https://mlsanity.com/paper/2103.15573/) (2021)
- [UltraPose: Synthesizing Dense Pose with 1 Billion Points by Human-body Decoupling 3D Model](https://mlsanity.com/paper/2110.15267/) (2021)
- [DensePose 3D: Lifting Canonical Surface Maps of Articulated Objects to the Third Dimension](https://mlsanity.com/paper/2109.00033/) (2021)
- [TemporalUV: Capturing Loose Clothing with Temporally Coherent UV Coordinates](https://mlsanity.com/paper/2204.03671/) (2022)
- [KTN: Knowledge Transfer Network for Learning Multi-person 2D-3D Correspondences](https://mlsanity.com/paper/2206.10090/) (2022)
- [Normal-guided Garment UV Prediction for Human Re-texturing](https://mlsanity.com/paper/2303.06504/) (2023)
- [Semi-supervised Dense Keypoints Using Unlabeled Multiview Images](https://mlsanity.com/paper/2109.09299/) (2021)
- [Pose Recognition in the Wild: Animal pose estimation using Agglomerative Clustering and Contrastive Learning](https://mlsanity.com/paper/2111.08259/) (2021)
- [Direct Dense Pose Estimation](https://mlsanity.com/paper/2204.01263/) (2022)
- [Human Pose-Constrained UV Map Estimation](https://mlsanity.com/paper/2501.08815/) (2025)

## Abstract (arXiv)

DensePose supersedes traditional landmark detectors by densely mapping image pixels to body surface coordinates. This power, however, comes at a greatly increased annotation time, as supervising the model requires to manually label hundreds of points per pose instance. In this work, we thus seek methods to significantly slim down the DensePose annotations, proposing more efficient data collection strategies. In particular, we demonstrate that if annotations are collected in video frames, their efficacy can be multiplied for free by using motion cues. To explore this idea, we introduce DensePose-Track, a dataset of videos where selected frames are annotated in the traditional DensePose manner. Then, building on geometric properties of the DensePose mapping, we use the video dynamic to propagate ground-truth annotations in time as well as to learn from Siamese equivariance constraints. Having performed exhaustive empirical evaluation of various data annotation and learning strategies, we demonstrate that doing so can deliver significantly improved pose estimation results over strong baselines. However, despite what is suggested by some recent works, we show that merely synthesizing motion patterns by applying geometric transformations to isolated frames is significantly less effective, and that motion cues help much more when they are extracted from videos.