• Stars
    star
    602
  • Rank 74,409 (Top 2 %)
  • Language
    Python
  • License
    Other
  • Created about 2 years ago
  • Updated about 1 year ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Fast-BEV: A Fast and Strong Bird’s-Eye View Perception Baseline

Fast-BEV

Fast-BEV: A Fast and Strong Bird’s-Eye View Perception Baseline image image image

Usage

usage

Installation

  • CUDA>=9.2
  • GCC>=5.4
  • Python>=3.6
  • Pytorch>=1.8.1
  • Torchvision>=0.9.1
  • MMCV-full==1.4.0
  • MMDetection==2.14.0
  • MMSegmentation==0.14.1

Dataset preparation

  .
  β”œβ”€β”€ data
  β”‚Β Β  └── nuscenes
  β”‚Β Β      β”œβ”€β”€ maps
  β”‚Β Β      β”œβ”€β”€ maps_bev_seg_gt_2class
  β”‚Β Β      β”œβ”€β”€ nuscenes_infos_test_4d_interval3_max60.pkl
  β”‚Β Β      β”œβ”€β”€ nuscenes_infos_train_4d_interval3_max60.pkl
  β”‚Β Β      β”œβ”€β”€ nuscenes_infos_val_4d_interval3_max60.pkl
  β”‚Β Β      β”œβ”€β”€ v1.0-test
  β”‚Β Β      └── v1.0-trainval

download

Pretraining

  .
  β”œβ”€β”€ pretrained_models
  β”‚   β”œβ”€β”€ cascade_mask_rcnn_r18_fpn_coco-mstrain_3x_20e_nuim_bbox_mAP_0.5110_segm_mAP_0.4070.pth
  β”‚   β”œβ”€β”€ cascade_mask_rcnn_r34_fpn_coco-mstrain_3x_20e_nuim_bbox_mAP_0.5190_segm_mAP_0.4140.pth
  β”‚   └── cascade_mask_rcnn_r50_fpn_coco-mstrain_3x_20e_nuim_bbox_mAP_0.5400_segm_mAP_0.4300.pth

download

Training

  .
  β”œβ”€β”€ work_dirs
    └── fastbev
      └── exp
          └── paper
              └── fastbev_m0_r18_s256x704_v200x200x4_c192_d2_f4
              β”‚   β”œβ”€β”€ epoch_20.pth
              β”‚   β”œβ”€β”€ latest.pth -> epoch_20.pth
              β”‚   β”œβ”€β”€ log.eval.fastbev_m0_r18_s256x704_v200x200x4_c192_d2_f4.02062323.txt
              β”‚   └── log.test.fastbev_m0_r18_s256x704_v200x200x4_c192_d2_f4.02062309.txt
              β”œβ”€β”€ fastbev_m1_r18_s320x880_v200x200x4_c192_d2_f4
              β”‚   β”œβ”€β”€ epoch_20.pth
              β”‚   β”œβ”€β”€ latest.pth -> epoch_20.pth
              β”‚   β”œβ”€β”€ log.eval.fastbev_m1_r18_s320x880_v200x200x4_c192_d2_f4.02080000.txt
              β”‚   └── log.test.fastbev_m1_r18_s320x880_v200x200x4_c192_d2_f4.02072346.txt
              β”œβ”€β”€ fastbev_m2_r34_s256x704_v200x200x4_c224_d4_f4
              β”‚   β”œβ”€β”€ epoch_20.pth
              β”‚   β”œβ”€β”€ latest.pth -> epoch_20.pth
              β”‚   β”œβ”€β”€ log.eval.fastbev_m2_r34_s256x704_v200x200x4_c224_d4_f4.02080021.txt
              β”‚   └── log.test.fastbev_m2_r34_s256x704_v200x200x4_c224_d4_f4.02080005.txt
              β”œβ”€β”€ fastbev_m4_r50_s320x880_v250x250x6_c256_d6_f4
              β”‚   β”œβ”€β”€ epoch_20.pth
              β”‚   β”œβ”€β”€ latest.pth -> epoch_20.pth
              β”‚   β”œβ”€β”€ log.eval.fastbev_m4_r50_s320x880_v250x250x6_c256_d6_f4.02080021.txt
              β”‚   └── log.test.fastbev_m4_r50_s320x880_v250x250x6_c256_d6_f4.02080005.txt
              └── fastbev_m5_r50_s512x1408_v250x250x6_c256_d6_f4
                  β”œβ”€β”€ epoch_20.pth
                  β”œβ”€β”€ latest.pth -> epoch_20.pth
                  β”œβ”€β”€ log.eval.fastbev_m5_r50_s512x1408_v250x250x6_c256_d6_f4.02080021.txt
                  └── log.test.fastbev_m5_r50_s512x1408_v250x250x6_c256_d6_f4.02080001.txt

download

Deployment

TODO

View Transformation Latency on device

2D-to-3D on CUDA & CPU

Citation

@article{li2023fast,
  title={Fast-BEV: A Fast and Strong Bird's-Eye View Perception Baseline},
  author={Li, Yangguang and Huang, Bin and Chen, Zeren and Cui, Yufeng and Liang, Feng and Shen, Mingzhu and Liu, Fenggang and Xie, Enze and Sheng, Lu and Ouyang, Wanli and others},
  journal={arXiv preprint arXiv:2301.12511},
  year={2023}
}