No such file or directory: '/media/raid5-array/experiments/Deep-PC/training_pose_errors_pytorch/kitti_pose_error_data_test_00.pickle

[BruceJKHuang]

when I run the following:

python test_dpc_net.py --seqs 00 --corr pose

it gives me errors.

However, I am pretty sure I change all the path variables in both the test_dpc_net.py and some in the create_kitti_training_data.py. I couldn't find any other paths that I need to change. Could you please look into it? Thanks.

Detected epoch 11 as the best one. Traceback (most recent call last): File "test_dpc_net.py", line 280, in main() File "test_dpc_net.py", line 81, in main test_stats, tm_dict = run_test(seq, epoch, correction_type, 'sparse', paths) File "test_dpc_net.py", line 168, in run_test test_loader = DataLoader(KITTIOdometryDataset(kitti_config['kitti_data_file'], img_type=train_config['img_type'], transform_img=transform_img, run_type='test', remap_kitti_folder=paths['kitti_img']), File "/home/brucebot/Cloud/workspace/dpc-net/utils.py", line 86, in init self.load_kitti_data(run_type) #Loads self.image_quad_paths and self.labels File "/home/brucebot/Cloud/workspace/dpc-net/utils.py", line 90, in load_kitti_data with open(self.pickle_file, 'rb') as handle: FileNotFoundError: [Errno 2] No such file or directory: '/media/raid5-array/experiments/Deep-PC/training_pose_errors_pytorch/kitti_pose_error_data_test_00.pickle'

ps.

I used grep and I found:

Binary file ./128.100.201.179/2017-dpc-net/trained_models/pose/seq_00_epoch_11_best.pth.tar matches Binary file ./128.100.201.179/2017-dpc-net/trained_models/pose/seq_05_epoch_11_best.pth.tar matches Binary file ./128.100.201.179/2017-dpc-net/trained_models/pose/seq_02_epoch_7_best.pth.tar matches Binary file ./128.100.201.179/2017-dpc-net/kitti_training_data/kitti_pose_error_data_test_05.pickle matches Binary file ./128.100.201.179/2017-dpc-net/kitti_training_data/kitti_pose_error_data_test_00.pickle matches Binary file ./128.100.201.179/2017-dpc-net/kitti_training_data/kitti_pose_error_data_test_02.pickle matches

[HyuanTan]

@BruceJKHuang Hi, when you load the pre-trained model such as seq_00_epoch_11_best.pth.tar, the saved_data is dictionary-like struct which have key kitti_config, and kitti_config also is dictionary-like struct which have key kitti_data_file:

    saved_data = torch.load(kitti_test_trained_model_path)
    train_config = saved_data['train_config']
    system_config = saved_data['system_config']
    kitti_config = saved_data['kitti_config']
    train_config["trained_models_dir"] = os.path.join(paths["trained_models"], train_config["correction_type"])
    kitti_data_file = kitti_config["kitti_data_file"].split("/")[-1]
    kitti_config["kitti_data_file"] = os.path.join(paths["kitti_vo"], kitti_data_file)

test_loader = DataLoader(KITTIOdometryDataset(kitti_config['kitti_data_file'], img_type=train_config['img_type'], transform_img=transform_img, run_type='test', remap_kitti_folder=paths['kitti_img']), 
                        batch_size=32,
                        shuffle=False, num_workers=6)

The DataLoader load the data using the path in kitti_config['kitti_data_file'], you can print the saved_data to confirm that. So you can change kitti_config['kitti_data_file'] to your own way. Hope that can help you.