Hi, I'm very interested in this excellent project. I wonder how you get a third-view hand detector? Is the faster rcnn trained on a dataset that only contains hands? I found that the hands cannot be detected in the following two pictures using the given demo and the pretrained model.
Hi, I'm very interested in this excellent project. I wonder how you get a third-view hand detector? Is the faster rcnn trained on a dataset that only contains hands? I found that the hands cannot be detected in the following two pictures using the given demo and the pretrained model.
![mmexport1652838142730](https://user-images.githubusercontent.com/43207078/168940105-0a018091-4ed4-4d3f-921d-116b882c4cb2.jpg)