guangxinghan / qa-fewdet Goto Github PK

Hello, author. However, I encountered this problem while running the code. When I run this command：sh /home/wenqian/PycharmProjects/QA-FewDet-main/scripts/meta_training_pascalvoc_split1_resnet101.sh
The problem is:
"RuntimeError: The size of tensor a (4) must match the size of tensor b (7) at non-singleton dimension 3."
I tried many methods, but couldn't solve them. How should I solve such a problem?

Command: sh scripts/meta_training_pascalvoc_split1_resnet101.sh

ValueError: Milestone must be smaller than total number of updates: num_updates=10000, milestone=10000

version: 0.5

cfg File:
SOLVER:
IMS_PER_BATCH: 4
BASE_LR: 0.002
STEPS: (15000, 20000)
MAX_ITER: 20000
CHECKPOINT_PERIOD: 10000

Has the author encountered this problem?

Environment:
Ubuntu18+torch1.8+cuda11.0 detetron2-v0.5

dear author, when I read the code, I found that when you construct the inter-class subgraph, you only update by calculating the similarity of each class prototype, without using the described GCN, and I just found an update for the base class category node, do not find the novel class node. Maybe it is my fault for not finding it, but can you help me? thank you very much!!!

Dear author，I'm debugging your model recently. I'm confused about some parameters in defaults.py. They are the following parameters：
_C.INPUT.FS.FEW_SHOT = False
_C.INPUT.BASE_CLASS_MEMORY= 60
_C.INPUT.FS.SUPPORT_EXCLUDE_QUERY = False

and why _C.DATASETS.TRAIN_KEEPCLASSES = 'all', _C.DATASETS.TEST_KEEPCLASSES = ' '
Is that mean try on all classes but test only on base classes?
Could you please help me，thanks a lot.

Hello, your work is very interesting. When I read the code, I found the following differences between QA-FewDet and fewx :
in fewx
bg_num_0 = max(1, min(fg_inds.shape[0] * 2, int(num_instances * 0.25))),
real_bg_topk_inds_0 = real_bg_inds[real_bg_inds < int(num_instances * 0.5)][:bg_num_0],
real_bg_topk_inds_1 = real_bg_inds[real_bg_inds >= int(num_instances * 0.5)][:bg_num_1]

in QA-FewD
bg_num_0 = max(1, min(fg_inds.shape[0] * 2, int(128 * 0.5)))
real_bg_topk_inds_0 = real_bg_inds[real_bg_inds < 128][:bg_num_0]
real_bg_topk_inds_1 = real_bg_inds[real_bg_inds >= 128][:bg_num_1]

According to my understanding，the first 128 proposals are negative support proposals, and the last 128 proposals are positive support proposals，this means that there is only one query image to participate in the training each time?
Could you please explain why you did this，thanks a lot.

Hi,

Great work!
I am trying to create a demo script but I am not familiar with these libraries.
How do I initialize the model with random support images?

Thanks in advance!
William

Hello author! I would like to ask, when I ran the code for Fewx using a graphics card with 4 24g memory (which is the baseline you used), I found that in the 10 shot fine-tuning of Pascalvoc split 1, nAP50 was 63.7%, which is significantly different from the 58.6% given in your paper. What is the reason for this?

Here is the result of my run:
[11/26 12:53:47] d2.evaluation.evaluator INFO: Inference done 1224/1238. 0.9081 s / img. ETA=0:00:12
[11/26 12:53:52] d2.evaluation.evaluator INFO: Inference done 1232/1238. 0.9063 s / img. ETA=0:00:05
[11/26 12:53:56] d2.evaluation.evaluator INFO: Total inference time: 0:18:43.905460 (0.911521 s / img per device, on 4 devices)
[11/26 12:53:56] d2.evaluation.evaluator INFO: Total inference pure compute time: 0:18:35 (0.904586 s / img per device, on 4 devices)
[11/26 12:54:51] d2.evaluation.testing INFO: copypaste: Task: bbox
[11/26 12:54:51] d2.evaluation.testing INFO: copypaste: AP,AP50,AP75,bAP,bAP50,bAP75,nAP,nAP50,nAP75
[11/26 12:54:51] d2.evaluation.testing INFO: copypaste: 39.1364,69.6365,40.0869,40.2169,71.6110,41.1615,35.8949,63.7130,36.8634
[11/26 12:54:51] d2.utils.events INFO: iter: 0 total_loss: 0.108 loss_cls: 0.026 loss_box_reg: 0.040 loss_rpn_cls: 0.034 loss_rpn_loc: 0.008 data_time: 21.6117 lr: 0.000001 max_mem: 6831M
[11/26 12:54:51] d2.engine.hooks INFO: Total training time: 0:20:06 (0:20:06 on hooks)

The configuration file for meta training is the same as the one you provided.

In the process of my meta-training, the inference process of k=1, 2, 3, 5, 10 was repeated twice, and the final AP value was exactly the same.Whether the problem about code?

What should the file path under datasets/coco/ be, for example, whether the train2014 and val2014 data folders are separate or together (trainval2014)? Thanks!

Hello, the coco dataset cannot reproduce the performance of the paper, but the voc dataset can. Is this because my coco data set has an error?