CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flow
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913584166993920 |
|---|---|
| author | Pan, Chenbin Yaman, Burhaneddin Velipasalar, Senem Ren, Liu |
| author_facet | Pan, Chenbin Yaman, Burhaneddin Velipasalar, Senem Ren, Liu |
| contents | Autonomous driving stands as a pivotal domain in computer vision, shaping the future of transportation. Within this paradigm, the backbone of the system plays a crucial role in interpreting the complex environment. However, a notable challenge has been the loss of clear supervision when it comes to Bird's Eye View elements. To address this limitation, we introduce CLIP-BEVFormer, a novel approach that leverages the power of contrastive learning techniques to enhance the multi-view image-derived BEV backbones with ground truth information flow. We conduct extensive experiments on the challenging nuScenes dataset and showcase significant and consistent improvements over the SOTA. Specifically, CLIP-BEVFormer achieves an impressive 8.5\% and 9.2\% enhancement in terms of NDS and mAP, respectively, over the previous best BEV model on the 3D object detection task. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_08919 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flow Pan, Chenbin Yaman, Burhaneddin Velipasalar, Senem Ren, Liu Computer Vision and Pattern Recognition Autonomous driving stands as a pivotal domain in computer vision, shaping the future of transportation. Within this paradigm, the backbone of the system plays a crucial role in interpreting the complex environment. However, a notable challenge has been the loss of clear supervision when it comes to Bird's Eye View elements. To address this limitation, we introduce CLIP-BEVFormer, a novel approach that leverages the power of contrastive learning techniques to enhance the multi-view image-derived BEV backbones with ground truth information flow. We conduct extensive experiments on the challenging nuScenes dataset and showcase significant and consistent improvements over the SOTA. Specifically, CLIP-BEVFormer achieves an impressive 8.5\% and 9.2\% enhancement in terms of NDS and mAP, respectively, over the previous best BEV model on the 3D object detection task. |
| title | CLIP-BEVFormer: Enhancing Multi-View Image-Based BEV Detector with Ground Truth Flow |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2403.08919 |