guods

About load bird.pth.tar

1

When I test the model bird.pth.tar, Missing key(s) in state_dict: "Conv2d_1a_3x3.conv.weight", "Conv2d_1a_3x3.bn.weight", .................

How long time is one imgae detected on your GPU? What is your GPU?

我用提换VGG16替换特征提取网络， 训练检测器时有正常的结果，但是当用训练好的VGG16检测器微调vgg16 step1时，没有检测框也没有分割结果，你可以猜测下什么原因吗

Low GPU utilization ratio

1

For Windows and 2060, GPU utilization is very low when it do inference.

For windows, I create the int8 engine with calibration data on one Tesla T4 , the size of the engine file is about 320M, but the speed of detection is...

the size of engine file

5

I create the int8 engine file on GPU 1060, the size of engine file is only 60M, the detection result is very poor. But I create the int8 engine file...

Multiple GPU

8

Thank you for your work, but I have some questions： How does the generated engine run on multiple graphics cards? How to set GPU Id number ？

How much memory of GPU is needed for inference? I test images with GTX 1060 (6G), error: out of memory!

If I want to train one or two class of coco data, how do I it?

the pretrained model

3

Would you share the pretrained model?