Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

detection result #12

Closed
mountain111 opened this issue Oct 21, 2021 · 6 comments
Closed

detection result #12

mountain111 opened this issue Oct 21, 2021 · 6 comments

Comments

@mountain111
Copy link

mountain111 commented Oct 21, 2021 •
edited

Applying PVT detection framework, I tried a CycleMLP-B1 based detector with RetinaNet 1x.
I got AP=27.1, fairly inferior to the reported 38.6. Could you give some advices to reproduce the reported result?

The specific configure is as follows

base = [
'base/models/retinanet_r50_fpn.py',
'base/datasets/coco_detection.py',
'base/schedules/schedule_1x.py',
'base/default_runtime.py'
]
#optimizer
model = dict(
pretrained='./pretrained/CycleMLP_B1.pth',
backbone=dict(
type='CycleMLP_B1_feat',
style='pytorch'),
neck=dict(
type='FPN',
in_channels=[64, 128, 320, 512],
out_channels=256,
start_level=1,
add_extra_convs='on_input',
num_outs=5))
#optimizer
optimizer = dict(delete=True, type='AdamW', lr=0.0001, weight_decay=0.0001)
optimizer_config = dict(grad_clip=None)

find_unused_parameters = True
@ShoufaChen
Copy link
Owner

Hi, thanks for your intestest.

Would you mind copy-pasting your training log (above the first itereation)?

@mountain111
Copy link
Author

mountain111 commented Oct 21, 2021 •
edited

The log is too long, and I pase the log for the first epoch.

2021-10-20 19:13:26,408 - mmdet - INFO - Epoch [1][50/7330] lr: 9.890e-06, eta: 1 day, 16:17:48, time: 1.650, data_time: 0.880, memory: 3951, loss_cls: 1.2268, loss_bbox: 0.6944, loss: 1.9211
2021-10-20 19:14:02,538 - mmdet - INFO - Epoch [1][100/7330] lr: 1.988e-05, eta: 1 day, 4:57:13, time: 0.723, data_time: 0.008, memory: 3951, loss_cls: 1.1547, loss_bbox: 0.6733, loss: 1.8280
2021-10-20 19:14:37,397 - mmdet - INFO - Epoch [1][150/7330] lr: 2.987e-05, eta: 1 day, 0:57:37, time: 0.697, data_time: 0.008, memory: 3952, loss_cls: 1.0479, loss_bbox: 0.6645, loss: 1.7124
2021-10-20 19:15:14,558 - mmdet - INFO - Epoch [1][200/7330] lr: 3.986e-05, eta: 23:14:21, time: 0.743, data_time: 0.008, memory: 3952, loss_cls: 0.9869, loss_bbox: 0.6628, loss: 1.6496
2021-10-20 19:15:50,191 - mmdet - INFO - Epoch [1][250/7330] lr: 4.985e-05, eta: 22:03:11, time: 0.713, data_time: 0.009, memory: 3952, loss_cls: 0.9701, loss_bbox: 0.6605, loss: 1.6306
2021-10-20 19:16:26,490 - mmdet - INFO - Epoch [1][300/7330] lr: 5.984e-05, eta: 21:18:45, time: 0.726, data_time: 0.009, memory: 3952, loss_cls: 0.9503, loss_bbox: 0.6423, loss: 1.5925
2021-10-20 19:17:03,185 - mmdet - INFO - Epoch [1][350/7330] lr: 6.983e-05, eta: 20:48:33, time: 0.734, data_time: 0.008, memory: 3952, loss_cls: 0.9458, loss_bbox: 0.6275, loss: 1.5733
2021-10-20 19:17:38,656 - mmdet - INFO - Epoch [1][400/7330] lr: 7.982e-05, eta: 20:21:15, time: 0.709, data_time: 0.009, memory: 3952, loss_cls: 0.9432, loss_bbox: 0.6319, loss: 1.5751
2021-10-20 19:18:14,900 - mmdet - INFO - Epoch [1][450/7330] lr: 8.981e-05, eta: 20:02:27, time: 0.725, data_time: 0.008, memory: 3952, loss_cls: 0.9315, loss_bbox: 0.6020, loss: 1.5335
2021-10-20 19:18:49,972 - mmdet - INFO - Epoch [1][500/7330] lr: 9.980e-05, eta: 19:43:49, time: 0.701, data_time: 0.009, memory: 3952, loss_cls: 0.9288, loss_bbox: 0.5965, loss: 1.5253
2021-10-20 19:19:26,109 - mmdet - INFO - Epoch [1][550/7330] lr: 1.000e-04, eta: 19:31:18, time: 0.723, data_time: 0.008,
2021-10-20 20:35:53,515 - mmdet - INFO - Epoch [1][6950/7330] lr: 1.000e-04, eta: 16:17:04, time: 0.759, data_time: 0.009, memory: 4169, loss_cls: 0.5645, loss_bbox: 0.4134, loss: 0.9778
2021-10-20 20:36:28,805 - mmdet - INFO - Exp name: retinanet_cycle_b1_fpn_1x_coco.py
2021-10-20 20:36:28,806 - mmdet - INFO - Epoch [1][7000/7330] lr: 1.000e-04, eta: 16:16:18, time: 0.706, data_time: 0.009, memory: 4169, loss_cls: 0.5908, loss_bbox: 0.4267, loss: 1.0175
2021-10-20 20:37:04,711 - mmdet - INFO - Epoch [1][7050/7330] lr: 1.000e-04, eta: 16:15:39, time: 0.718, data_time: 0.009, memory: 4169, loss_cls: 0.5868, loss_bbox: 0.4170, loss: 1.0038
2021-10-20 20:37:40,085 - mmdet - INFO - Epoch [1][7100/7330] lr: 1.000e-04, eta: 16:14:53, time: 0.707, data_time: 0.010, memory: 4169, loss_cls: 0.5842, loss_bbox: 0.4152, loss: 0.9994
2021-10-20 20:38:15,594 - mmdet - INFO - Epoch [1][7150/7330] lr: 1.000e-04, eta: 16:14:10, time: 0.711, data_time: 0.009, memory: 4169, loss_cls: 0.5776, loss_bbox: 0.4209, loss: 0.9985
2021-10-20 20:38:51,236 - mmdet - INFO - Epoch [1][7200/7330] lr: 1.000e-04, eta: 16:13:28, time: 0.713, data_time: 0.008, memory: 4169, loss_cls: 0.5820, loss_bbox: 0.4121, loss: 0.9941
2021-10-20 20:39:26,102 - mmdet - INFO - Epoch [1][7250/7330] lr: 1.000e-04, eta: 16:12:37, time: 0.697, data_time: 0.008, memory: 4169, loss_cls: 0.6032, loss_bbox: 0.4200, loss: 1.0232
2021-10-20 20:40:01,991 - mmdet - INFO - Epoch [1][7300/7330] lr: 1.000e-04, eta: 16:11:58, time: 0.718, data_time: 0.008, memory: 4169, loss_cls: 0.5791, loss_bbox: 0.4121, loss: 0.9913
2021-10-20 20:40:33,915 - mmdet - INFO - Saving checkpoint at 1 epochs

DONE (t=16.74s).
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.065
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=1000 ] = 0.123
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=1000 ] = 0.061
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=1000 ] = 0.028
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=1000 ] = 0.068
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=1000 ] = 0.099
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.231
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=300 ] = 0.231
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=1000 ] = 0.231
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=1000 ] = 0.086
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=1000 ] = 0.223
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=1000 ] = 0.366

@ShoufaChen
Copy link
Owner

I am sorry but I need the log information about loading the pretrained model, which is not included.

@mountain111
Copy link
Author

Thanks a lot for your reply!
Previously, the pretrained model was not loaded. Now, I add an initialization method for pretrained models.
Hope the result can be improved.

@nikky4D
Copy link

nikky4D commented Oct 22, 2021

Thanks a lot for your reply! Previously, the pretrained model was not loaded. Now, I add an initialization method for pretrained models. Hope the result can be improved.

Were the results improved in your training

@mountain111
Copy link
Author

mountain111 commented Oct 24, 2021 •
edited

Thanks a lot for your reply! Previously, the pretrained model was not loaded. Now, I add an initialization method for pretrained models. Hope the result can be improved.

Were the results improved in your training

The AP has been improved to 38.6, the same as the reported value.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@nikky4D @ShoufaChen @mountain111