Eric Sheng

icon company @NVIDIA icon location China icon location AI Computing

Results 13 comments of Eric Sheng

Built a WeChat group

![qr](https://user-images.githubusercontent.com/62684483/85361856-98462880-b54f-11ea-8df4-292fc0416c5b.jpg)

Built a WeChat group

![Screenshot_20200713_124142](https://user-images.githubusercontent.com/62684483/87271693-9516da80-c506-11ea-9d23-03528edf4306.jpg)

Built a WeChat group

![mmqrcode1595413748507](https://user-images.githubusercontent.com/62684483/88166664-34944580-cc4a-11ea-80d3-da890e7cb2fc.png)

Built a WeChat group

![Screenshot_20200729_145045](https://user-images.githubusercontent.com/62684483/88767786-afaaae00-d1ac-11ea-80d2-f76efe673f7b.jpg)

Built a WeChat group

![Screenshot_20200810_105345](https://user-images.githubusercontent.com/62684483/89748962-5faae000-daf8-11ea-8bc9-a2c1c567150c.jpg)

Built a WeChat group

![Screenshot_20200819_100224](https://user-images.githubusercontent.com/62684483/90584618-84443d80-e205-11ea-9e62-0f07926b9c9d.jpg)

Did you check accuracy of YOLOv4 by using your implementation on MSCOCO?

This is result on COCO validation set (val2017), input is 416*416. It seems it does not reach optimal accuracy of the original DarkNet model ``` conf_thresh = 0.001 NMS_IOU_thresh =...

Did you check accuracy of YOLOv4 by using your implementation on MSCOCO?

> @ersheng-ai > > * Did you use **converted** or **trained** weights on Pytorch? > * Do you use keeping aspect ration of image during resizing? > YOLOv4 doesn't keep...

Did you check accuracy of YOLOv4 by using your implementation on MSCOCO?

> @ersheng-ai Try to use resizing without keeping-aspect-ratio/padding-zeros, will accuracy be better? I will try it

Did you check accuracy of YOLOv4 by using your implementation on MSCOCO?

Zero padding is removed so that images of any ratio are squashed or stretched to 416 * 416 Now the results look much closer to https://github.com/AlexeyAB/darknet/issues/5354 ``` Average Precision (AP)...