Add files via upload

Signed-off-by: Hao Lu <[email protected]>

tools/Deepaccident_converter.py

@@ -0,0 +1,168 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import pickle
+
+import numpy as np
+from nuscenes import NuScenes
+from nuscenes.utils.data_classes import Box
+from pyquaternion import Quaternion
+import mmcv
+
+
+cam_names = ['Camera_FrontLeft', 'Camera_Front', 'Camera_FrontRight', 'Camera_BackLeft', 'Camera_Back', 'Camera_BackRight']
+
+map_name_from_general_to_detection = {
+    'car':'car',
+    'truck':'truck',
+    'van':'truck',
+    'bus':'bus',
+    'emergency_vehicle':'ignore',
+    'other_vehicle':'car',
+    'motorcycle':'motorcycle',
+    'bicycle':'bicycle',
+    'cyclist':'bicycle',
+    'pedestrian':'pedestrian',
+    'animal':'ignore',
+    'human.pedestrian.adult': 'pedestrian',
+    'human.pedestrian.child': 'pedestrian',
+    'human.pedestrian.wheelchair': 'ignore',
+    'human.pedestrian.stroller': 'ignore',
+    'human.pedestrian.personal_mobility': 'ignore',
+    'human.pedestrian.police_officer': 'pedestrian',
+    'human.pedestrian.construction_worker': 'pedestrian',
+    'animal': 'ignore',
+    'vehicle.car': 'car',
+    'vehicle.motorcycle': 'motorcycle',
+    'vehicle.bicycle': 'bicycle',
+    'vehicle.bus.bendy': 'bus',
+    'vehicle.bus.rigid': 'bus',
+    'vehicle.truck': 'truck',
+    'vehicle.construction': 'construction_vehicle',
+    'vehicle.emergency.ambulance': 'ignore',
+    'vehicle.emergency.police': 'ignore',
+    'vehicle.trailer': 'trailer',
+    'movable_object.barrier': 'barrier',
+    'movable_object.trafficcone': 'traffic_cone',
+    'movable_object.pushable_pullable': 'ignore',
+    'movable_object.debris': 'ignore',
+    'static_object.bicycle_rack': 'ignore',
+}
+# classes = [
+#     'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
+#     'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
+# ]
+classes = [
+    'car', 'truck', 'van',
+    'motorcycle', 'bicycle', 'pedestrian'
+] # label 的顺序0-5
+
+
+def get_gt(info):
+    """Generate gt labels from info.
+    Args:
+        info(dict): Infos needed to generate gt labels.
+
+    Returns:
+        Tensor: GT bboxes.
+        Tensor: GT labels.
+    """
+    ego2global_rotation = info['cams']['CAM_FRONT']['ego2global_rotation']
+    ego2global_translation = info['cams']['CAM_FRONT'][
+        'ego2global_translation']
+    trans = -np.array(ego2global_translation)
+    rot = Quaternion(ego2global_rotation).inverse
+    gt_boxes = list()
+    gt_labels = list()
+
+    for ann_info in info['ann_infos']:
+        # Use ego coordinate.
+        if (map_name_from_general_to_detection[ann_info['category_name']]
+                not in classes): # or ann_info['num_lidar_pts'] + ann_info['num_radar_pts'] <= 0
+            print('*********************************')
+            print(ann_info['category_name'])
+            continue
+        box = Box(
+            ann_info['translation'],
+            ann_info['size'],
+            Quaternion(ann_info['rotation']),
+            velocity=ann_info['velocity'],
+        )
+        box.translate(trans)
+        box.rotate(rot)
+        box_xyz = np.array(box.center)
+        box_dxdydz = np.array(box.wlh)[[1, 0, 2]]
+        box_yaw = np.array([box.orientation.yaw_pitch_roll[0]])
+        box_velo = np.array(box.velocity[:2])
+        print('box_xyz',box_xyz)
+        print('box_dxdydz',box_dxdydz)
+        print('box_yaw',box_yaw)
+        print('box_velo',box_velo)
+        gt_box = np.concatenate([box_xyz, box_dxdydz, box_yaw, box_velo])
+        gt_boxes.append(gt_box)
+
+        gt_labels.append(
+            classes.index(
+                map_name_from_general_to_detection[ann_info['category_name']]))
+    return gt_boxes, gt_labels
+
+
+
+def get_gt_DeepAccident2nus(info):
+    gt_boxes = list()
+    gt_labels = list()
+    for id in range(len(info['gt_boxes'])):
+        if info['camera_visibility'][id]==1:
+            temp_box = info['gt_boxes'][id]
+            temp_box[1] = -1.0 * temp_box[1]
+            temp_box[2] = temp_box[2] + 1.9
+            gt_boxes.append(np.concatenate([info['gt_boxes'][id], info['gt_velocity'][id]]))
+            gt_labels.append(
+                classes.index(
+                    map_name_from_general_to_detection[info['gt_names'][id]]))
+    # print(info['gt_names'])
+    return gt_boxes, gt_labels
+
+def add_ann_adj_info(dataset, save_path, dataroot='./data/DeepAccident/'):
+    for id in range(len(dataset['infos'])):
+        if id % 10 == 0:
+            print('%d/%d' % (id, len(dataset['infos'])))
+        info = dataset['infos'][id]
+        # get sweep adjacent frame info
+        dataset['infos'][id]['ann_infos'] = get_gt_DeepAccident2nus(info)
+        dataset['infos'][id]['scene_token'] = info['scenario_type'] + info['vehicle_name'] +  info['scene_name']
+        # get sensor2ego_translation
+        for cam_name in cam_names:
+            cam_infos = dataset['infos'][id]['cams'][cam_name]
+            lidar2cam_rt = cam_infos['lidar_to_camera_matrix']
+            ego2lidar_rt = np.linalg.inv(dataset['infos'][id]['lidar_to_ego_matrix'])
+            ego2cam = lidar2cam_rt @ ego2lidar_rt
+            ego2cam_rt = ego2cam.T
+            ego2cam_rt = ego2cam_rt[[1, 2, 0], 0:3]
+            ego2cam_rt[1, :] = -ego2cam_rt[1, :]
+            ego2cam_rt[0, :] = -ego2cam_rt[0, :]
+            lidar_to_camera_matrix = ego2cam_rt
+            temp = np.array(cam_infos['lidar_to_camera_matrix'][0:3, 3][[1, 2, 0]])
+            temp[1] = -1.0 * temp[1]
+            view = np.eye(4)
+            view[:3, :3] = np.array(lidar_to_camera_matrix)
+            view[:3, -1] = temp
+            ego2img = np.linalg.inv(view)
+            ego2im_Q = Quaternion(matrix=np.array(ego2img[0:3, 0:3]))
+            sensor2ego_rotation = [ego2im_Q.w.astype(float), ego2im_Q.x.astype(float), ego2im_Q.y.astype(float),
+                                   ego2im_Q.z.astype(float)]
+            dataset['infos'][id]['cams'][cam_name]['sensor2ego_rotation'] = sensor2ego_rotation
+            dataset['infos'][id]['cams'][cam_name]['sensor2ego_translation'] = (ego2img[0:3, -1] + np.array([0.0,0.0,1.9])).tolist()
+    with open(save_path,
+              'wb') as fid:
+        pickle.dump(dataset, fid)
+
+
+
+if __name__ == '__main__':
+    DeepAccident_pkl_path = '/mnt/cfs/algorithm/hao.lu/Code/BEVDepth_pg/data/DeepAccident_data/carla_infos_train.pkl'
+    DeepAccident_data = mmcv.load(DeepAccident_pkl_path)
+    add_ann_adj_info(dataset=DeepAccident_data, save_path='/mnt/cfs/algorithm/hao.lu/Code/BEVDepth_pg/data/DeepAccident_data/bevdetv4-DeepAccident_infos_train.pkl')
+
+    DeepAccident_pkl_path = '/mnt/cfs/algorithm/hao.lu/Code/BEVDepth_pg/data/DeepAccident_data/carla_infos_val.pkl'
+    DeepAccident_data = mmcv.load(DeepAccident_pkl_path)
+    add_ann_adj_info(dataset=DeepAccident_data, save_path='/mnt/cfs/algorithm/hao.lu/Code/BEVDepth_pg/data/DeepAccident_data/bevdetv4-DeepAccident_infos_val.pkl')
+#

tools/Lyft_converter.py

@@ -0,0 +1,194 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import pickle
+
+import numpy as np
+from nuscenes import NuScenes
+from nuscenes.utils.data_classes import Box
+from pyquaternion import Quaternion
+from lyft_dataset_sdk import LyftDataset
+from tools.data_converter import nuscenes_converter as nuscenes_converter
+
+map_name_from_general_to_detection = {
+    'car':'car',
+    'truck':'truck',
+    'bus':'bus',
+    'emergency_vehicle':'ignore',
+    'other_vehicle':'car',
+    'motorcycle':'motorcycle',
+    'bicycle':'bicycle',
+    'pedestrian':'pedestrian',
+    'animal':'ignore',
+    'human.pedestrian.adult': 'pedestrian',
+    'human.pedestrian.child': 'pedestrian',
+    'human.pedestrian.wheelchair': 'ignore',
+    'human.pedestrian.stroller': 'ignore',
+    'human.pedestrian.personal_mobility': 'ignore',
+    'human.pedestrian.police_officer': 'pedestrian',
+    'human.pedestrian.construction_worker': 'pedestrian',
+    'animal': 'ignore',
+    'vehicle.car': 'car',
+    'vehicle.motorcycle': 'motorcycle',
+    'vehicle.bicycle': 'bicycle',
+    'vehicle.bus.bendy': 'bus',
+    'vehicle.bus.rigid': 'bus',
+    'vehicle.truck': 'truck',
+    'vehicle.construction': 'construction_vehicle',
+    'vehicle.emergency.ambulance': 'ignore',
+    'vehicle.emergency.police': 'ignore',
+    'vehicle.trailer': 'trailer',
+    'movable_object.barrier': 'barrier',
+    'movable_object.trafficcone': 'traffic_cone',
+    'movable_object.pushable_pullable': 'ignore',
+    'movable_object.debris': 'ignore',
+    'static_object.bicycle_rack': 'ignore',
+}
+classes = [
+    'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
+    'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
+]
+
+def get_gt(info):
+    """Generate gt labels from info.
+
+    Args:
+        info(dict): Infos needed to generate gt labels.
+
+    Returns:
+        Tensor: GT bboxes.
+        Tensor: GT labels.
+    """
+    ego2global_rotation = info['cams']['CAM_FRONT']['ego2global_rotation']
+    ego2global_translation = info['cams']['CAM_FRONT'][
+        'ego2global_translation']
+    trans = -np.array(ego2global_translation)
+    rot = Quaternion(ego2global_rotation).inverse
+    gt_boxes = list()
+    gt_labels = list()
+
+    for ann_info in info['ann_infos']:
+        # Use ego coordinate.
+        if (map_name_from_general_to_detection[ann_info['category_name']]
+                not in classes): # or ann_info['num_lidar_pts'] + ann_info['num_radar_pts'] <= 0
+            print('*********************************')
+            print(ann_info['category_name'])
+            continue
+        box = Box(
+            ann_info['translation'],
+            ann_info['size'],
+            Quaternion(ann_info['rotation']),
+            velocity=ann_info['velocity'],
+        )
+        box.translate(trans)
+        box.rotate(rot)
+        box_xyz = np.array(box.center)
+        box_dxdydz = np.array(box.wlh)[[1, 0, 2]]
+        box_yaw = np.array([box.orientation.yaw_pitch_roll[0]])
+        box_velo = np.array(box.velocity[:2])
+        print('box_xyz',box_xyz)
+        print('box_dxdydz',box_dxdydz)
+        print('box_yaw',box_yaw)
+        print('box_velo',box_velo)
+        gt_box = np.concatenate([box_xyz, box_dxdydz, box_yaw, box_velo])
+        gt_boxes.append(gt_box)
+
+        gt_labels.append(
+            classes.index(
+                map_name_from_general_to_detection[ann_info['category_name']]))
+    return gt_boxes, z
+
+
+
+def add_ann_adj_info(extra_tag):
+    nuscenes_version = 'v1.0-trainval'
+    dataroot = './data/lyft/'
+    for set in ['train', 'test']:
+        dataset = pickle.load(
+            open('./data/lyft/%s_infos_%s.pkl' % (extra_tag, set), 'rb'))
+        nuscenes = LyftDataset(dataroot, dataroot + 'v1.01-' + set + '/v1.01-'+set)
+        for id in range(len(dataset['infos'])):
+            if id % 10 == 0:
+                print('%d/%d' % (id, len(dataset['infos'])))
+            info = dataset['infos'][id]
+            # get sweep adjacent frame info
+            sample = nuscenes.get('sample', info['token'])
+            ann_infos = list()
+            for ann in sample['anns']:
+                ann_info = nuscenes.get('sample_annotation', ann)
+                velocity = nuscenes.box_velocity(ann_info['token'])
+                if np.any(np.isnan(velocity)):
+                    velocity = np.zeros(3)
+                ann_info['velocity'] = velocity
+                ann_infos.append(ann_info)
+            dataset['infos'][id]['ann_infos'] = ann_infos
+            dataset['infos'][id]['ann_infos'] = get_gt(dataset['infos'][id])
+            dataset['infos'][id]['scene_token'] = sample['scene_token']
+        with open('./data/lyft/%s_infos_ann_%s.pkl' % (extra_tag, set),
+                  'wb') as fid:
+            pickle.dump(dataset, fid)
+
+
+
+def add_ann_adj_info_val(extra_tag):
+    dataroot = './data/lyft/'
+    set = 'val'
+    dataset = pickle.load(
+        open('./data/lyft/%s_infos_%s.pkl' % (extra_tag, set), 'rb'))
+    nuscenes = LyftDataset(dataroot, dataroot + 'v1.01-train' + '/v1.01-train')
+    for id in range(len(dataset['infos'])):
+        if id % 10 == 0:
+            print('%d/%d' % (id, len(dataset['infos'])))
+        info = dataset['infos'][id]
+        # get sweep adjacent frame info
+        sample = nuscenes.get('sample', info['token'])
+        ann_infos = list()
+        for ann in sample['anns']:
+            ann_info = nuscenes.get('sample_annotation', ann)
+            velocity = nuscenes.box_velocity(ann_info['token'])
+            if np.any(np.isnan(velocity)):
+                velocity = np.zeros(3)
+            ann_info['velocity'] = velocity
+            ann_infos.append(ann_info)
+        dataset['infos'][id]['ann_infos'] = ann_infos
+        dataset['infos'][id]['ann_infos'] = get_gt(dataset['infos'][id])
+        dataset['infos'][id]['scene_token'] = sample['scene_token']
+    with open('./data/lyft/%s_infos_ann_%s.pkl' % (extra_tag, set),
+              'wb') as fid:
+        pickle.dump(dataset, fid)
+
+
+if __name__ == '__main__':
+
+    dataset = 'lyft'
+    version = 'v1.01'
+    train_version = f'{version}-trainval'
+    root_path = './data/nuscenes'
+    extra_tag = 'lyft'
+    print('add_ann_infos')
+    add_ann_adj_info(extra_tag)
+    add_ann_adj_info_val(extra_tag)
+
+
+#
+# dataroot = './data/lyft/'
+# set = 'train'
+# dataset = pickle.load(
+#     open('./data/lyft/%s_infos_%s.pkl' % (extra_tag, set), 'rb'))
+# nuscenes = LyftDataset(dataroot, dataroot + 'v1.01-' + set + '/v1.01-' + set)
+# for id in range(len(dataset['infos'])):
+#     if id % 10 == 0:
+#         print('%d/%d' % (id, len(dataset['infos'])))
+#     info = dataset['infos'][id]
+#     # get sweep adjacent frame info
+#     sample = nuscenes.get('sample', info['token'])
+#     ann_infos = list()
+#     for ann in sample['anns']:
+#         ann_info = nuscenes.get('sample_annotation', ann)
+#         velocity = nuscenes.box_velocity(ann_info['token'])
+#         if np.any(np.isnan(velocity)):
+#             velocity = np.zeros(3)
+#         ann_info['velocity'] = velocity
+#         ann_infos.append(ann_info)
+#     dataset['infos'][id]['ann_infos'] = ann_infos
+#     dataset['infos'][id]['ann_infos'] = get_gt(dataset['infos'][id])
+#     print(dataset['infos'][id]['ann_infos'])
+#     dataset['infos'][id]['scene_token'] = sample['scene_token']