zhangzhengde0225 / cdnet Goto Github PK

Hi @zhangzhengde,

I'm trying to download the trainet but I'm having a problem. Since I don't live in China, I can't download from Baidu Pan site - it requires a Chinese phone number (so mine doesn't work), I tried to use Pandownload but the file is too big and can't be downloaded with this site. Also, the Jbox site seems to be down (https://jbox.sjtu.edu.cn/l/Z0i6nQ).

Would it be possible that you provide a link to download the trainset on another platform such as Mega or Mediafire?
Thank you for you project :)

      模型保存时是按那个指标的？`fi = fitness(np.array(results).reshape(1, -1))  # fitness_i = weighted combination of [P, R, mAP, F1]`

您好，我看过你的论文，觉得你的工作非常好。其中我注意到，你的工作中只关注了F1 score这一个指标，恰好我的工作中也使用到这个指标，所以想问问你一个问题。 yolov5中保存模型会计算一个最优值，计算的是map0.5和map的加权和
fi = fitness(np.array(results).reshape(1, -1)
w = [0.0, 0.0, 0.1, 0.9] # weights for [P, R, [email protected], [email protected]:0.95]
我想请问下你的工作保存模型是以那个指标来进行保存的，我看到你的注释中写道
# weighted combination of [P, R, mAP, F1]
请问是map和F1的加权和吗，权重比例是多少，是0.1 * map+0.9 * F1吗？ 还是只取最大的F1 score？

Hey！From the experimental results, we can see that SSVM greatly improves the detection accuracy，Although the flow of the algorithm is described in the text, I still do not quite understand the principle and implementation details of the algorithm.
Can you briefly describe how SSVM works? And can you recommend some reference material on SSVM?

您好，请问能开源数据集转coco格式的代码么，或者coco格式的数据集网盘，yolo2coco.py代码好像不是完整的，import damei as dm找不到在哪

Is the marking of the fig error? Should it be SE-YOLOv5 + NST + SSVM？

您好，按照论文中的设置训练，在testsets_1770数据集做测试，没有达到论文中的f1 score，论文中有95%，复现只有85%；fps比论文中快了很多，fps在40-55之间；请问是我忽略了什么吗

hi，the green line which is means use se-yolov5+nst is likely better than the pink line which is just use se-yolov5+nst?

在运行detect.py时一直遇到这个问题，cv2.error: OpenCV(4.5.4) D:\a\opencv-python\opencv-python\opencv\modules\imgproc\src\resize.cpp:4051: error: (-215:Assertion failed) !ssize.empty() in function 'cv::resize'没有改动过您的代码，想问一下原因，谢谢您

can you please update dataset link

您好，非常感谢您所做的工作！
我在我的仓库 CDNet-yolov5 正在尝试利用您的数据集训练一些常用的检测模型，这项工作现在正在进行，不知道能否融入到您的 CDNet仓库的 README.md中，这项工作是完全基于现有的Yolov5的V6.1版本。

Can you provide the corresponding model download link

您在论文中写道“在训练过程中，将同时检测人行横道和引导箭头，引导箭头将从最终的检测结果中排除。这样实际上增加了训练中导箭的损失函数。”，请问您是怎么将引导箭头将从最终的检测结果中排除的？我在您的detect.py代码中并没有看到相应操作，您能指点我一下吗？

may you upload models to google drive, i can't register to Baidu since i dont have Chinese number

我想做一个车载相机视角下的数据集，但是网上很多图片和视频有水印，很难去获取到，你们能告诉我一些渠道去获取这些图片或者视频吗？十分感谢