VL-PLM

VL-PLM copied to clipboard

_pickle.UnpicklingError

1

If I set num-gpu to 4 and try to train Mask R-CNN with Large Scale Jitter, I'm going to say, _pickle.UnpicklingError: pickle data was truncated An error occurs. How can...

th0214

Accuracy with classes

5

Hi, I often see 'Giraffe' or other wild animals appear around ~70% when inferencing on images. I would just like to confirm I am using the correct class set (COCO_CATEGORIES...

xsacha

PL generation categories

3

Hi thanks for providing such a nice work! I have a question about the categories that you've used for PL generation. It seems that you used only novel COCO categories...

jihwanp

您好，进行训练必须在多gpu的环境下嘛

3

minC0819

Download link seems to be broken⚡

6

**Issue Description:** I encountered an issue while trying to download the pre-trained weights—the download link seems to be broken. **Request:** Could you please provide a new link for downloading the...

LinSY546749

我通过实验发现CLIP对目标检测的裁剪图像进行判别效果并不好，CLIP更加关注粗粒度的目标，例如他会将一个吃热狗的人给予热狗更大的概率，而实际标签是人。请问你有遇到这种问题吗？有什么解决方法吗？

nhw649

The code for the semi-supervised experiment

4

Where are the code and model for the semi-supervised experiment?

fcjian

代码训练问题，权重最优，图像融合

4

您好！ 我看了您的代码，想请问您下面两个问题： 1. 请问您是如何保存效果最优的那一轮权重的？代码里好像只保存了最后一轮的权重，但似乎有时候前面的权重会比后面的权重测试效果更好。 2. 请问为何您的图像融合部分不放在训练代码中，直接端到端的输出，而要放在测试代码里，有什么理由么？ 期待您的回复。

guigui6666

How to train my own dataset ？

1

Your work is outstanding, how can I train my own dataset, which includes categories outside the COCO 80 categories.

0x0f0f0f0f

Errors occurred while training and evaluating the model During training: TypeError:__ init__ () missing 2 required keyword-only arguments: 'bg_ cls_ loss_ weight' and 'openset_ test' When profiling: KeyError: 'naiveSyncBN_ N'

3

During training： Traceback (most recent call last): File "train_net.py", line 246, in args=(args,), File "/media/cheng/dataset4/annaconda3/envs/clip/lib/python3.7/site-packages/detectron2-RegionCLIP-py3.7-linux-x86_64.egg/detectron2/engine/launch.py", line 82, in launch main_func(*args) File "train_net.py", line 203, in main model = build_model(cfg) File...

Originlightwkp