# Towards Learning to Imitate from a Single Video Demonstration

arXiv: https://arxiv.org/abs/1901.07186 · HTML: https://mlsanity.com/paper/1901.07186/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 22 Jan 2019.

## Built on by (most cited first, 7 of 7)

- [Visual Adversarial Imitation Learning using Variational Models](https://mlsanity.com/paper/2107.08829/) (2021)
- [FLD: Fourier Latent Dynamics for Structured Motion Representation and Learning](https://mlsanity.com/paper/2402.13820/) (2024)
- [A Survey of Demonstration Learning](https://mlsanity.com/paper/2303.11191/) (2023)
- [Contrastive Learning from Demonstrations](https://mlsanity.com/paper/2201.12813/) (2022)
- [Imitation from Observation With Bootstrapped Contrastive Learning](https://mlsanity.com/paper/2302.06540/) (2023)
- [Imitation Learning of Correlated Policies in Stackelberg Games](https://mlsanity.com/paper/2503.08883/) (2025)
- [Learning Quadruped Walking from Seconds of Demonstration](https://mlsanity.com/paper/2603.06961/) (2026)

## Abstract (arXiv)

Agents that can learn to imitate given video observation -- without direct access to state or action information are more applicable to learning in the natural world. However, formulating a reinforcement learning (RL) agent that facilitates this goal remains a significant challenge. We approach this challenge using contrastive training to learn a reward function comparing an agent's behaviour with a single demonstration. We use a Siamese recurrent neural network architecture to learn rewards in space and time between motion clips while training an RL policy to minimize this distance. Through experimentation, we also find that the inclusion of multi-task data and additional image encoding losses improve the temporal consistency of the learned rewards and, as a result, significantly improves policy learning. We demonstrate our approach on simulated humanoid, dog, and raptor agents in 2D and a quadruped and a humanoid in 3D. We show that our method outperforms current state-of-the-art techniques in these environments and can learn to imitate from a single video demonstration.