| import numpy as np |
| import os |
| import cv2 |
|
|
| def calculate_2d_projections(coordinates_3d, intrinsics): |
| """ |
| Input: |
| coordinates: [3, N] |
| intrinsics: [3, 3] |
| Return |
| projected_coordinates: [N, 2] |
| """ |
| projected_coordinates = intrinsics @ coordinates_3d |
| projected_coordinates = projected_coordinates[:2, :] / projected_coordinates[2, :] |
| projected_coordinates = projected_coordinates.T |
| projected_coordinates = np.array(projected_coordinates, dtype=np.int32) |
|
|
| return projected_coordinates |
|
|
| def get_3d_bbox(scale, shift = 0): |
| """ |
| Input: |
| scale: [3] or scalar |
| shift: [3] or scalar |
| Return |
| bbox_3d: [3, N] |
| |
| """ |
| if hasattr(scale, "__iter__"): |
| bbox_3d = np.array([[scale[0] / 2, +scale[1] / 2, scale[2] / 2], |
| [scale[0] / 2, +scale[1] / 2, -scale[2] / 2], |
| [-scale[0] / 2, +scale[1] / 2, scale[2] / 2], |
| [-scale[0] / 2, +scale[1] / 2, -scale[2] / 2], |
| [+scale[0] / 2, -scale[1] / 2, scale[2] / 2], |
| [+scale[0] / 2, -scale[1] / 2, -scale[2] / 2], |
| [-scale[0] / 2, -scale[1] / 2, scale[2] / 2], |
| [-scale[0] / 2, -scale[1] / 2, -scale[2] / 2]]) + shift |
| else: |
| bbox_3d = np.array([[scale / 2, +scale / 2, scale / 2], |
| [scale / 2, +scale / 2, -scale / 2], |
| [-scale / 2, +scale / 2, scale / 2], |
| [-scale / 2, +scale / 2, -scale / 2], |
| [+scale / 2, -scale / 2, scale / 2], |
| [+scale / 2, -scale / 2, -scale / 2], |
| [-scale / 2, -scale / 2, scale / 2], |
| [-scale / 2, -scale / 2, -scale / 2]]) +shift |
|
|
| bbox_3d = bbox_3d.transpose() |
| return bbox_3d |
|
|
| def draw_3d_bbox(img, imgpts, color, size=1): |
| imgpts = np.int32(imgpts).reshape(-1, 2) |
|
|
| |
| color_ground = (int(color[0] * 0.3), int(color[1] * 0.3), int(color[2] * 0.3)) |
| for i, j in zip([4, 5, 6, 7],[5, 7, 4, 6]): |
| img = cv2.line(img, tuple(imgpts[i]), tuple(imgpts[j]), color_ground, size) |
|
|
| |
| color_pillar = (int(color[0]*0.6), int(color[1]*0.6), int(color[2]*0.6)) |
| for i, j in zip(range(4),range(4,8)): |
| img = cv2.line(img, tuple(imgpts[i]), tuple(imgpts[j]), color_pillar, size) |
|
|
| |
| for i, j in zip([0, 1, 2, 3],[1, 3, 0, 2]): |
| img = cv2.line(img, tuple(imgpts[i]), tuple(imgpts[j]), color, size) |
| return img |
|
|
| def draw_3d_pts(img, imgpts, color, size=1): |
| imgpts = np.int32(imgpts).reshape(-1, 2) |
| for point in imgpts: |
| img = cv2.circle(img, (point[0], point[1]), size, color, -1) |
| return img |
|
|
| def draw_detections(image, pred_rots, pred_trans, model_points, intrinsics, color=(255, 0, 0)): |
| num_pred_instances = len(pred_rots) |
| draw_image_bbox = image.copy() |
| |
| scale = (np.max(model_points, axis=0) - np.min(model_points, axis=0)) |
| shift = np.mean(model_points, axis=0) |
| print(scale, shift) |
|
|
| bbox_3d = get_3d_bbox(scale, shift) |
|
|
| |
| choose = np.random.choice(np.arange(len(model_points)), 512) |
| pts_3d = model_points[choose].T |
|
|
| for ind in range(num_pred_instances): |
| |
| transformed_bbox_3d = pred_rots[ind]@bbox_3d + pred_trans[ind][:,np.newaxis] |
| projected_bbox = calculate_2d_projections(transformed_bbox_3d, intrinsics[ind]) |
| draw_image_bbox = draw_3d_bbox(draw_image_bbox, projected_bbox, color) |
| |
| transformed_pts_3d = pred_rots[ind]@pts_3d + pred_trans[ind][:,np.newaxis] |
| projected_pts = calculate_2d_projections(transformed_pts_3d, intrinsics[ind]) |
| draw_image_bbox = draw_3d_pts(draw_image_bbox, projected_pts, color) |
|
|
| return draw_image_bbox |
|
|
