starsky68
starsky68
How to customize and implement bidirectional RNN
我只使用了swin transformer的模型架构，win size是4，输入大小是32，其他都没变，也没有使用任何数据增强，resnet可以达到80%的acc，但是swin trans只有52%，请问，这样的结果，是什么引起的？
How to prune a custom tensor? The tensor is a recursive variable and is initialized with tf.zeros.
Prior to filing: check that this should be a bug instead of a feature request. Everything supported, including the compatible versions of TensorFlow, is listed in the overview page of...
Engine was exported successfully, but using the detection of the project, the detection effect is very poor, and even most frames are wrong. What is the problem in that link
训练自己的图像会出现下面的问题： UserWarning: Corrupt EXIF data. Expecting to read 12 bytes but only got 10.
训练的reid模型可以转成onnx吗
我在运行yolov3_deepsort_eval.py的时候，法务得到正常的结果，我替换了yolo5的检测器。 with VideoTracker(cfg, args, video_path, result_filename) as vdo_trk: vdo_trk.run() 该段代码只是使用了yolov3_deepsort.py中的追踪，没有任何返回值嘛？这这个调用了mot的数据了嘛？ 可以对yolov3_deepsort_eval.py进行一下详细的说明嘛。谢谢
间隔3，5，7，9帧，对deepsort的影响，有参数可以调节嘛？
请问，该项目有转换为多类别多目标跟踪的版本吗？
（pytorch >= 1.1 and torchvision >= 0.3）When compiling the build.sh file in the nms directory, the following error occurred: Permission denied: '/usr/local/lib/python3.8/dist-packages/test-easy-install-2225652.write-test'