# Towards Consistent Long-Term Pose Generation

arXiv: https://arxiv.org/abs/2507.18382 · HTML: https://mlsanity.com/paper/2507.18382/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 30 of its 37 references that are in the map.
- First posted on arXiv 24 Jul 2025.

## Builds on (most cited first, 30 of 30)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [Human Motion Diffusion Model](https://mlsanity.com/paper/2209.14916/) (2022)
- [Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields](https://mlsanity.com/paper/1611.08050/) (2016)
- [What Matters in Learning from Offline Human Demonstrations for Robot Manipulation](https://mlsanity.com/paper/2108.03298/) (2021)
- [Autoregressive Image Generation without Vector Quantization](https://mlsanity.com/paper/2406.11838/) (2024)
- [MediaPipe: A Framework for Building Perception Pipelines](https://mlsanity.com/paper/1906.08172/) (2019)
- [Faith and Fate: Limits of Transformers on Compositionality](https://mlsanity.com/paper/2305.18654/) (2023)
- [Action-Conditioned 3D Human Motion Synthesis with Transformer VAE](https://mlsanity.com/paper/2104.05670/) (2021)
- [Action2Motion: Conditioned Generation of 3D Human Motions](https://mlsanity.com/paper/2007.15240/) (2020)
- [MoMask: Generative Masked Modeling of 3D Human Motions](https://mlsanity.com/paper/2312.00063/) (2023)
- [TM2T: Stochastic and Tokenized Modeling for the Reciprocal Generation of 3D Human Motions and Texts](https://mlsanity.com/paper/2207.01696/) (2022)
- [Language2Pose: Natural Language Grounded Pose Forecasting](https://mlsanity.com/paper/1907.01108/) (2019)
- [Diffusion-based Generation, Optimization, and Planning in 3D Scenes](https://mlsanity.com/paper/2301.06015/) (2023)
- [First-Person Hand Action Benchmark with RGB-D Videos and 3D Hand Pose Annotations](https://mlsanity.com/paper/1704.02463/) (2017)
- [Long-term Human Motion Prediction with Scene Context](https://mlsanity.com/paper/2007.03672/) (2020)
- [The pitfalls of next-token prediction](https://mlsanity.com/paper/2403.06963/) (2024)
- [HUMANISE: Language-conditioned Human Motion Generation in 3D Scenes](https://mlsanity.com/paper/2210.09729/) (2022)
- [Synthesizing Long-Term 3D Human Motion and Interaction in 3D Scenes](https://mlsanity.com/paper/2012.05522/) (2020)
- [HumanMAC: Masked Motion Completion for Human Motion Prediction](https://mlsanity.com/paper/2302.03665/) (2023)
- [Move as You Say, Interact as You Can: Language-guided Human Motion Generation with Scene Affordance](https://mlsanity.com/paper/2403.18036/) (2024)
- [TM2D: Bimodality Driven 3D Dance Generation via Music-Text Integration](https://mlsanity.com/paper/2304.02419/) (2023)
- [ChatPose: Chatting about 3D Human Pose](https://mlsanity.com/paper/2311.18836/) (2023)
- [Action-Agnostic Human Pose Forecasting](https://mlsanity.com/paper/1810.09676/) (2018)
- [Predicting 3D Human Dynamics from Video](https://mlsanity.com/paper/1908.04781/) (2019)
- [GenHowTo: Learning to Generate Actions and State Transformations from Instructional Videos](https://mlsanity.com/paper/2312.07322/) (2023)
- [LEGO: Learning EGOcentric Action Frame Generation via Visual Instruction Tuning](https://mlsanity.com/paper/2312.03849/) (2023)
- [Forecasting Human Dynamics from Static Images](https://mlsanity.com/paper/1704.03432/) (2017)
- [Contact-aware Human Motion Generation from Textual Descriptions](https://mlsanity.com/paper/2403.15709/) (2024)

## Built on by (most cited first, 1 of 1)

- [Mistake Attribution: Fine-Grained Mistake Understanding in Egocentric Videos](https://mlsanity.com/paper/2511.20525/) (2025)

## Abstract (arXiv)

Current approaches to pose generation rely heavily on intermediate representations, either through two-stage pipelines with quantization or autoregressive models that accumulate errors during inference. This fundamental limitation leads to degraded performance, particularly in long-term pose generation where maintaining temporal coherence is crucial. We propose a novel one-stage architecture that directly generates poses in continuous coordinate space from minimal context - a single RGB image and text description - while maintaining consistent distributions between training and inference. Our key innovation is eliminating the need for intermediate representations or token-based generation by operating directly on pose coordinates through a relative movement prediction mechanism that preserves spatial relationships, and a unified placeholder token approach that enables single-forward generation with identical behavior during training and inference. Through extensive experiments on Penn Action and First-Person Hand Action Benchmark (F-PHAB) datasets, we demonstrate that our approach significantly outperforms existing quantization-based and autoregressive methods, especially in long-term generation scenarios.