Dear author, thank you very much for your work. I have a question and hope to get your answer.
I first tested your pre-training model to achieve the mAP effect you described. However, when training cadDn.yaml, batCHsize =1 and epoch=10 were set. But the trained model, mAP=0. Could you give me some advice?
AP:0.0000, 0. 0000，0. 0000
Cyclist APQ0.50， 0.50， 0.50:
bbox AP:0.0000， 0.0000， 0.0000
bev AP : 0.0000, 0.0000，0. 0000
3dAP:0. 0000，0.0000， 0. 0000
Cyclist AP R40@0.50， 0.50，0.50:
bbox AP:0.0000， 0.0000， 0.0000
bev AP:0.0000， 0.0000， 0.0000
3d AP :0.0000, 0.0000， 0.0000
Cyclist APQ0.50， 0.25， 0.25:
bbox AP :0.0000, 0.0000， 0.0000
bev AP:0.0000, 0.0000， 0.0000
3dAP:O.0000，0. 0000，0. 0000
cyclist AP R40@0.50, 0.25，0.25:
bbox AP:O.0000， 0.0000， 0. 0000
bev AP:0.0000， 0.0000， 0.0000
3dAP:0.0000，0. 0000，0. 0000
2021-11- 30 11 : 30:56, 482
INFO Result is save to /data01/ zq/CaDDN/ output/ kitti models/CaDDN/ default/eval/eval
with_ train/epoch _2/val
2021-11-30 11:30:56，482
INFO ** **** Evaluation done.
2021-11- 30 11 : 30 :56, 507
INFO Epoch 2 has been evaluated
2021- 11-30 11:30:56，508
INFO
==> Loading parameters from checkpoint /data01/ zq/ CaDDN/ output/ kitti models/CaD
DN/ def ault/ckpt/checkpoint epoch 3.pth to GPU
2021- 11-30 11:31:05, 034
INFO ==> Checkpoint trained from version: pcdet+0 .3.0+0000000
2021-11-30 11:31:11,871
INFO
==> Done ( Loaded 903/ 903 )
2021-11-30 11:31:11, 897
INFO
***** EPOCH 3 EVALUATION
eval:
0%|
0/3769 [00:00<?, ?it/s]
/home/ omnisky/ zq/ tib/python3. 6/ site- packages/ torch/nn/ functional. py :2705: UserWarning: Default grid sample and
affine grid behavior has changed to align corners=False since 1.3.0. Please specify align corners-True if the c
ld behavior is desired. See the documentation of grid sample for details.
warnings .warn("Default grid sample and affine_ grid behavior has changed
eval: 24%|2| 911/3769 [05:06<16:36， 2.87it/s, recal
Dear author, thank you very much for your work. I have a question and hope to get your answer. I first tested your pre-training model to achieve the mAP effect you described. However, when training cadDn.yaml, batCHsize =1 and epoch=10 were set. But the trained model, mAP=0. Could you give me some advice? AP:0.0000, 0. 0000，0. 0000 Cyclist APQ0.50， 0.50， 0.50: bbox AP:0.0000， 0.0000， 0.0000 bev AP : 0.0000, 0.0000，0. 0000 3dAP:0. 0000，0.0000， 0. 0000 Cyclist AP R40@0.50， 0.50，0.50: bbox AP:0.0000， 0.0000， 0.0000 bev AP:0.0000， 0.0000， 0.0000 3d AP :0.0000, 0.0000， 0.0000 Cyclist APQ0.50， 0.25， 0.25: bbox AP :0.0000, 0.0000， 0.0000 bev AP:0.0000, 0.0000， 0.0000 3dAP:O.0000，0. 0000，0. 0000 cyclist AP R40@0.50, 0.25，0.25: bbox AP:O.0000， 0.0000， 0. 0000 bev AP:0.0000， 0.0000， 0.0000 3dAP:0.0000，0. 0000，0. 0000 2021-11- 30 11 : 30:56, 482 INFO Result is save to /data01/ zq/CaDDN/ output/ kitti models/CaDDN/ default/eval/eval with_ train/epoch _2/val 2021-11-30 11:30:56，482 INFO ** **** Evaluation done.
2021-11- 30 11 : 30 :56, 507 INFO Epoch 2 has been evaluated 2021- 11-30 11:30:56，508 INFO ==> Loading parameters from checkpoint /data01/ zq/ CaDDN/ output/ kitti models/CaD DN/ def ault/ckpt/checkpoint epoch 3.pth to GPU 2021- 11-30 11:31:05, 034 INFO ==> Checkpoint trained from version: pcdet+0 .3.0+0000000 2021-11-30 11:31:11,871 INFO ==> Done ( Loaded 903/ 903 ) 2021-11-30 11:31:11, 897 INFO ***** EPOCH 3 EVALUATION eval: 0%| 0/3769 [00:00<?, ?it/s] /home/ omnisky/ zq/ tib/python3. 6/ site- packages/ torch/nn/ functional. py :2705: UserWarning: Default grid sample and affine grid behavior has changed to align corners=False since 1.3.0. Please specify align corners-True if the c ld behavior is desired. See the documentation of grid sample for details. warnings .warn("Default grid sample and affine_ grid behavior has changed eval: 24%|2| 911/3769 [05:06<16:36， 2.87it/s, recal