why “get_binimg” Donot crop and resize

[wuzuowuyou]

Randomly crop and resize image data， why get_binimg Donot crop and resize

https://github.com/nv-tlabs/lift-splat-shoot/blob/d74598cb51101e2143097ab270726a561f81f8fd/src/data.py#L171C19-L171C19

def get_binimg(self, rec):
        egopose = self.nusc.get('ego_pose',
                                self.nusc.get('sample_data', rec['data']['LIDAR_TOP'])['ego_pose_token'])
        trans = -np.array(egopose['translation'])
        rot = Quaternion(egopose['rotation']).inverse
        img = np.zeros((self.nx[0], self.nx[1]))
        for tok in rec['anns']:
            inst = self.nusc.get('sample_annotation', tok)
            # add category for lyft
            if not inst['category_name'].split('.')[0] == 'vehicle':
                continue
            box = Box(inst['translation'], inst['size'], Quaternion(inst['rotation']))
            box.translate(trans)
            box.rotate(rot)

            pts = box.bottom_corners()[:2].T
            pts = np.round(
                (pts - self.bx[:2] + self.dx[:2]/2.) / self.dx[:2]
                ).astype(np.int32)
            pts[:, [1, 0]] = pts[:, [0, 1]]
            cv2.fillPoly(img, [pts], 1.0)

        return torch.Tensor(img).unsqueeze(0)

[manueldiaz96]

Because this is the BEV representation, not a camera image. So it represents the space around the vehicle