GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks
We propose a novel approach for generating high-quality, synthetic data for domain-specific learning tasks, for which training data may not be readily available. We leverage recent progress in image-to-image translation to bridge the gap between simulated and real images, allowing us to generate rea...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Institute of Electrical and Electronics Engineers (IEEE)
2020
|
| Online Access: | https://hdl.handle.net/1721.1/125861 |
| _version_ | 1826206524294299648 |
|---|---|
| author | Stein, Gregory Joseph Roy, Nicholas |
| author2 | Massachusetts Institute of Technology. Department of Aeronautics and Astronautics |
| author_facet | Massachusetts Institute of Technology. Department of Aeronautics and Astronautics Stein, Gregory Joseph Roy, Nicholas |
| author_sort | Stein, Gregory Joseph |
| collection | MIT |
| description | We propose a novel approach for generating high-quality, synthetic data for domain-specific learning tasks, for which training data may not be readily available. We leverage recent progress in image-to-image translation to bridge the gap between simulated and real images, allowing us to generate realistic training data for real-world tasks using only unlabeled real-world images and a simulation. GeneSIS-Rtameliorates the burden of having to collect labeled real-world images and is a promising candidate for generating high-quality, domain-specific, synthetic data. To show the effectiveness of using GeneSIS-Rtto create training data, we study two tasks: semantic segmentation and reactive obstacle avoidance. We demonstrate that learning algorithms trained using data generated by GeneSIS-RT make high-accuracy predictions and outperform systems trained on raw simulated data alone, and as well or better than those trained on real data. Finally, we use our data to train a quadcopter to fly 60 meters at speeds up to 3.4 m/s through a cluttered environment, demonstrating that our GeneSIS-RT images can be used to learn to perform mission-critical tasks. |
| first_indexed | 2024-09-23T13:34:18Z |
| format | Article |
| id | mit-1721.1/125861 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T13:34:18Z |
| publishDate | 2020 |
| publisher | Institute of Electrical and Electronics Engineers (IEEE) |
| record_format | dspace |
| spelling | mit-1721.1/1258612022-10-01T15:44:14Z GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks Stein, Gregory Joseph Roy, Nicholas Massachusetts Institute of Technology. Department of Aeronautics and Astronautics Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science We propose a novel approach for generating high-quality, synthetic data for domain-specific learning tasks, for which training data may not be readily available. We leverage recent progress in image-to-image translation to bridge the gap between simulated and real images, allowing us to generate realistic training data for real-world tasks using only unlabeled real-world images and a simulation. GeneSIS-Rtameliorates the burden of having to collect labeled real-world images and is a promising candidate for generating high-quality, domain-specific, synthetic data. To show the effectiveness of using GeneSIS-Rtto create training data, we study two tasks: semantic segmentation and reactive obstacle avoidance. We demonstrate that learning algorithms trained using data generated by GeneSIS-RT make high-accuracy predictions and outperform systems trained on raw simulated data alone, and as well or better than those trained on real data. Finally, we use our data to train a quadcopter to fly 60 meters at speeds up to 3.4 m/s through a cluttered environment, demonstrating that our GeneSIS-RT images can be used to learn to perform mission-critical tasks. Defense Advanced Research Project Agency (DARPA) (Contract HR0011-15-C-0110). 2020-06-18T13:57:41Z 2020-06-18T13:57:41Z 2018-09 2018-05 2019-10-31T13:22:29Z Article http://purl.org/eprint/type/ConferencePaper 9781538630815 2577-087X https://hdl.handle.net/1721.1/125861 Stein, Gregory J. and Nicholas Roy. "GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks, IEEE International Conference on Robotics and Automation (ICRA), May 2018, Brisbane, QLD, Australia, Institute of Electrical and Electronics Engineers, September 2018. © 2018 IEEE en http://dx.doi.org/10.1109/icra.2018.8462971 IEEE International Conference on Robotics and Automation (ICRA) Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Institute of Electrical and Electronics Engineers (IEEE) arXiv |
| spellingShingle | Stein, Gregory Joseph Roy, Nicholas GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks |
| title | GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks |
| title_full | GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks |
| title_fullStr | GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks |
| title_full_unstemmed | GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks |
| title_short | GeneSIS-Rt: Generating Synthetic Images for Training Secondary Real-World Tasks |
| title_sort | genesis rt generating synthetic images for training secondary real world tasks |
| url | https://hdl.handle.net/1721.1/125861 |
| work_keys_str_mv | AT steingregoryjoseph genesisrtgeneratingsyntheticimagesfortrainingsecondaryrealworldtasks AT roynicholas genesisrtgeneratingsyntheticimagesfortrainingsecondaryrealworldtasks |