Sense-GVT / Fast-BEV

Fast-BEV: A Fast and Strong Bird’s-Eye View Perception Baseline
Other
568 stars 85 forks source link

Pretrained model size mismatch #78

Open Terencedu opened 4 months ago

Terencedu commented 4 months ago

Hi team,

I trained the pretrained model with the setting below. Why would some warnings appear? How to solve it?

dataset: v1.0-mini

pretrained model: cascade_mask_rcnn_r18_fpn_coco-mstrain_3x_20e_nuim_bbox_mAP_0.5110_segm_mAP_0.4070.pth

training code: python tools/train.py configs/fastbev/exp/paper/fastbev_m0_r18_s256x704_v200x200x4_c192_d2_f4.py --work-dir work_dir --gpu-ids 0

截图 2024-02-23 09-47-34 截图 2024-02-23 09-47-52

But these warnings do not affect the training of the model. Is the trained .pth file accurate? 截图 2024-02-23 09-48-03