# ArTIST: Autoregressive Trajectory Inpainting and Scoring for Tracking

arXiv: https://arxiv.org/abs/2004.07482 · HTML: https://mlsanity.com/paper/2004.07482/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Apr 2020.

## Built on by (most cited first, 6 of 6)

- [View Vertically: A Hierarchical Network for Trajectory Prediction via Fourier Spectrums](https://mlsanity.com/paper/2110.07288/) (2021)
- [MSN: Multi-Style Network for Trajectory Prediction](https://mlsanity.com/paper/2107.00932/) (2021)
- [Another Vertical View: A Hierarchical Network for Heterogeneous Trajectory Prediction via Spectrums](https://mlsanity.com/paper/2304.05106/) (2023)
- [SocialCircle+: Learning the Angle-based Conditioned Interaction Representation for Pedestrian Trajectory Prediction](https://mlsanity.com/paper/2409.14984/) (2024)
- [Resonance: Learning to Predict Social-Aware Pedestrian Trajectories as Co-Vibrations](https://mlsanity.com/paper/2412.02447/) (2024)
- [Socialality Anchors: Towards Group-bounded Trajectory Prediction](https://mlsanity.com/paper/2609.36852/) (2026)

## Abstract (arXiv)

One of the core components in online multiple object tracking (MOT) frameworks is associating new detections with existing tracklets, typically done via a scoring function. Despite the great advances in MOT, designing a reliable scoring function remains a challenge. In this paper, we introduce a probabilistic autoregressive generative model to score tracklet proposals by directly measuring the likelihood that a tracklet represents natural motion. One key property of our model is its ability to generate multiple likely futures of a tracklet given partial observations. This allows us to not only score tracklets but also effectively maintain existing tracklets when the detector fails to detect some objects even for a long time, e.g., due to occlusion, by sampling trajectories so as to inpaint the gaps caused by misdetection. Our experiments demonstrate the effectiveness of our approach to scoring and inpainting tracklets on several MOT benchmark datasets. We additionally show the generality of our generative model by using it to produce future representations in the challenging task of human motion prediction.