PeizeSun / SparseR-CNN

[CVPR2021, PAMI2023] End-to-End Object Detection with Learnable Proposal

PeizeSun/SparseR-CNN Issues

I am just start to learn, I found that the default configuration of Sparse R-CNN had better AP results than Faster R-CNN. However, with the detection results, I noticed that Sparse R-CNN had lower confidence scores for its detections. Why is that the case? im little confused
Updated 2 months ago
请问怎么使用预训练权重，从头开始训练的准确率比较低
Updated 4 months ago
我把从coco数据集换成了自己的数据集，但是出现了错误，我应该修改你的代码中的哪几个部分的代码
Updated 5 months ago5
How to visualize the activation maps of dynamic head in Sparse R-CNN？
Updated 9 months ago
"Dataset 'CrowdHuman_train' is not registered
Updated 9 months ago3
When set num_heads = 1, doesnt train
Updated 9 months ago12
Initialization box selection
Updated a year ago
Some questions about positional encoding, in the Table 10
Updated a year ago
The results were not reproduced on CrowdHuman dataset
Updated a year ago1
Init proposal box Visualization
Closed 3 years ago2
how to do Grid and Rondom proposal boxes initialization?
Updated 2 years ago5
KeyError: 1, when using only 1 class dataset to evaluation
Updated 2 years ago
AssertionError: Attribute 'thing_classes' in the metadata of 'coco_2017_train' cannot be set to a different value!
Updated 2 years ago2
About performance of Sparse R-CNN
Closed 2 years ago
Why the optimization targets of refined bboxes is assigned by refined box instead of unrefined bbox?
Closed 2 years ago2
检测不了物体
Closed 2 years ago1
关于动态头的级联问题
Closed 2 years ago
the learning rate adjustment when training with one gpu or two gpus
Updated 2 years ago5
Hi,Have you verified the model to ONNX, i meet some bugs to onnx.
Updated 2 years ago2
'assert (boxes1[:, 2:] >= boxes1[:, :2]).all()' happened when training
Updated 3 years ago4
How can I use a custom dataset?
Updated 3 years ago2
How to install the Sparse R-CNN(it seems amusing)?
Updated 3 years ago1
Change ADAMW to SGD
Closed 3 years ago4
Confusing about the performance in the paper
Closed 3 years ago3
How to do use test time augmentation?
Updated 3 years ago1
Does anything changes in detectron2？
Updated 3 years ago1
Resume started from zero again!
Updated 3 years ago3
Question about the proposal_boxes in the code
Closed 3 years ago2
How the feature re-use is performed
Updated 3 years ago1
how to visualize train log?
Closed 3 years ago
how to use plot function?
Updated 3 years ago1
Question about the evaluation
Closed 3 years ago10
保存模型与提供模型大小之间的差异
Closed 3 years ago1
Question about the Iterative architecture
Updated 3 years ago2
About the loss format
Updated 3 years ago4
train w/o pre-trained backbone
Updated 3 years ago3
question about pred_bboxes.detach() in DynamicHead.forward()
Updated 3 years ago2
Multi-head attention in dynamic head
Closed 3 years ago3
Question regarding proposal feature
Updated 3 years ago4
Will you release DetCo code and model version?
Closed 3 years ago3
Does input data annotations need to be normalized coordinate before training?
Closed 3 years ago2
issues about Mix-Precision training
Updated 3 years ago1
Have you tried smooth l1 + giou?
Closed 3 years ago1
question about learned proposal boxes
Updated 3 years ago1
cuda out of memory issue
Closed 3 years ago
在小数据集上for example （train:60 test:13）如何调优参数，使用默认配置测试结果不太好
Closed 4 years ago1
Randomness in the results for COCO
Closed 4 years ago2
how to detect mang images one time?
Closed 4 years ago1
how to change the confidence score in train net.py when eval-only?
Updated 4 years ago1
after trained about 7000 iters.it has error.
Updated 4 years ago2