preprint · arXiv (Cornell University) · 2019
Towards Object Detection from Motion
Rico Jonschkowski, Austin Stone
Cited by 2Open access
Abstract
We present a novel approach to weakly supervised object detection. Instead of annotated images, our method only requires two short videos to learn to detect a new object: 1) a video of a moving object and 2) one or more "negative" videos of the scene without the object. The key idea of our algorithm is to train the object detector to produce physically plausible object motion when applied to the first video and to not detect anything in the second video. With this approach, our method learns to locate objects without any object location annotations. Once the model is trained, it performs object detection on single images. We evaluate our method in three robotics settings that afford learning objects from motion: observing moving objects, watching demonstrations of object manipulation, and physically interacting with objects (see a video summary at https://youtu.be/BH0Hv3zZG_4).
Cite this paper
Jonschkowski, R., & Stone, A. (2019). Towards Object Detection from Motion. arXiv (Cornell University). https://doi.org/10.48550/arxiv.1909.12950
Jonschkowski, Rico, and Austin Stone. “Towards Object Detection from Motion.” arXiv (Cornell University), 2019, https://doi.org/10.48550/arxiv.1909.12950.
Jonschkowski, Rico, and Austin Stone. 2019. “Towards Object Detection from Motion.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.1909.12950.
Jonschkowski, R. and Stone, A. (2019) ‘Towards Object Detection from Motion’, arXiv (Cornell University). Available at: https://doi.org/10.48550/arxiv.1909.12950.
@misc{jonschkowski2019towards,
title = {{Towards Object Detection from Motion}},
author = {Jonschkowski, Rico and Stone, Austin},
journal = {arXiv (Cornell University)},
year = {2019},
doi = {10.48550/arxiv.1909.12950}
}Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.