search

mks0601 / 3DMPPE_POSENET_RELEASE

Official PyTorch implementation of "Camera Distance-aware Top-down Approach for 3D Multi-person Pose Estimation from a Single RGB Image", ICCV 2019
MIT License
807 stars 147 forks source link

Issue about bounding box value in your demo.py #88

Open xunhui-huang opened 3 years ago

xunhui-huang commented 3 years ago

Thanks for your excellent work!I have a problem about your provided bounding box value in your demo.py. bbox_list = [[139.41, 102.25, 222.39, 241.57],\ [287.17, 61.52, 74.88, 165.61],\ [540.04, 48.81, 99.96, 223.36],\ [372.58, 170.84, 266.63, 217.19],\ [0.5, 43.74, 90.1, 220.09]] # xmin, ymin, width, height However,I try to use these value to draw bounding box in the provided input.jpg,The location of bounding box is far from the human body.So how can i get the value of bbox_list,and whether the value of 'xmin' and 'ymin' refers to the center of bbox?

mks0601 commented 3 years ago

xmin and ymin means minimum x and y coordinates. In other word, top-left point of the box.

xunhui-huang commented 3 years ago

Thanks for your answer, I have another question. whether the demo.py has restrictions on the size of input image?I wish to implement your code to accept the wild image as input.Do I need to resize my input image?

mks0601 commented 3 years ago

If you set bbox, the code will automatically crop and resize the bbox area to 256x256 size and fed it to the network.

xunhui-huang commented 3 years ago

Thx.

sainan-zhang commented 2 years ago

Thanks for your wonderful work! I also have a problem about your provided bounding box value in your demo.py. How can I set th e value of bbox_list of my own input image? I found that bounding boxes are gotten from detect net in your paper. Should I use the the pre-trained ResNet50 to get it?

mks0601 commented 2 years ago

You can use any off-the-shelf human detector, such as detectron2 and yolov5.

sainan-zhang commented 2 years ago

Thanks very much!