Accelerating learned descriptor generation for visual localization
Visual SLAM systems use traditional feature extractor to retrieve features, a pair consisting of a keypoint and descriptor, from images. These features can then be matched to estimate the camera pose. However, these traditional feature extractors are surpassed by newer deep learning-based feature ex...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/175279 |
| _version_ | 1826112357594562560 |
|---|---|
| author | Liu, Woon Kit |
| author2 | Lam Siew Kei |
| author_facet | Lam Siew Kei Liu, Woon Kit |
| author_sort | Liu, Woon Kit |
| collection | NTU |
| description | Visual SLAM systems use traditional feature extractor to retrieve features, a pair consisting of a keypoint and descriptor, from images. These features can then be matched to estimate the camera pose. However, these traditional feature extractors are surpassed by newer deep learning-based feature extractor in the presence of imaging noise, illumination, or viewpoint changes. However, such AI models may suffer performance issues when deployed to embedded devices, which prioritises low-powered consumption. This report investigates the potential of deep learning accelerator libraries to accelerate feature extractor models for application in visual SLAM systems, particularly on embedded devices. TensorRT, is such a library that this can help achieve a significant speedup compared to traditional feature extraction methods. |
| first_indexed | 2024-10-01T03:05:41Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/175279 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T03:05:41Z |
| publishDate | 2024 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1752792024-04-26T15:44:00Z Accelerating learned descriptor generation for visual localization Liu, Woon Kit Lam Siew Kei School of Computer Science and Engineering ASSKLam@ntu.edu.sg Computer and Information Science Visual SLAM systems use traditional feature extractor to retrieve features, a pair consisting of a keypoint and descriptor, from images. These features can then be matched to estimate the camera pose. However, these traditional feature extractors are surpassed by newer deep learning-based feature extractor in the presence of imaging noise, illumination, or viewpoint changes. However, such AI models may suffer performance issues when deployed to embedded devices, which prioritises low-powered consumption. This report investigates the potential of deep learning accelerator libraries to accelerate feature extractor models for application in visual SLAM systems, particularly on embedded devices. TensorRT, is such a library that this can help achieve a significant speedup compared to traditional feature extraction methods. Bachelor's degree 2024-04-23T02:06:41Z 2024-04-23T02:06:41Z 2024 Final Year Project (FYP) Liu, W. K. (2024). Accelerating learned descriptor generation for visual localization. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/175279 https://hdl.handle.net/10356/175279 en SCSE23-0143 application/pdf Nanyang Technological University |
| spellingShingle | Computer and Information Science Liu, Woon Kit Accelerating learned descriptor generation for visual localization |
| title | Accelerating learned descriptor generation for visual localization |
| title_full | Accelerating learned descriptor generation for visual localization |
| title_fullStr | Accelerating learned descriptor generation for visual localization |
| title_full_unstemmed | Accelerating learned descriptor generation for visual localization |
| title_short | Accelerating learned descriptor generation for visual localization |
| title_sort | accelerating learned descriptor generation for visual localization |
| topic | Computer and Information Science |
| url | https://hdl.handle.net/10356/175279 |
| work_keys_str_mv | AT liuwoonkit acceleratinglearneddescriptorgenerationforvisuallocalization |