search

Cc-Hy / CMKD

Cross-Modality Knowledge Distillation Network for Monocular 3D Object Detection (ECCV 2022 Oral)
Apache License 2.0
107 stars 9 forks source link

Can not reproduce the paper's results. #53

Open hoiliu-0801 opened 11 months ago

hoiliu-0801 commented 11 months ago

Following the instructions, I trained BEV 60 epoch first, and then 20 epoch for detection. Here are the results I got. image

My training code is shown bellow : Step1: python train_cmkd.py --cfg cfgs/kitti_models/CMKD/CMKD-scd/cmkd_kitti_R50_scd_bev.yaml --pretrained_lidar_model ../checkpoints/scd-teacher-kitti.pth Step2: python test_cmkd.py --cfg cfgs/kitti_models/CMKD/CMKD-scd/cmkd_kitti_R50_scd_V2.yaml --ckpt ../output/kitti_models/CMKD/CMKD-scd/cmkd_kitti_R50_scd_bev/default/ckpt/checkpoint_epoch_60.pth Is anything wrong with the training pipeline? Please point it out for me, thank you.

Cc-Hy commented 8 months ago

Hi, have you tried distributed training?