search

happinesslz / EPNet

EPNet: Enhancing Point Features with Image Semantics for 3D Object Detection(ECCV 2020)
MIT License
234 stars 36 forks source link

About handling details #18

Closed ZacharyZgw closed 3 years ago

ZacharyZgw commented 3 years ago

Hello, i have a question about image size. SUNRGBD have different image size. How you deal with them. Can you tell me the details. Thank you very much!

XiwuChen commented 3 years ago

We set a larger size, i.e. (544, 768), to ensure that each image is not cropped. 

img_mask = np.zeros((544, 768, 3), dtype = np.float32)
shape = img.shape
img_mask[:shape[0], :shape[1], :] = img
ret_dict['img'] = img_mask.transpose((2, 0, 1)).astype(np.float32)

This operation is similar to https://github.com/happinesslz/EPNet/blob/0123c341243846aa3b412addcb9e2c07fd305237/lib/datasets/kitti_dataset.py#L54-L55, since images in the KITTI dataset also have different size.