autovision-cloud / ssl-lanes

[CoRL-2022] SSL-Lanes: Self-Supervised Learning for Motion Forecasting in Autonomous Driving

Hi there 👋

• 🌱 I'm Prarthana Bhattacharyya, currently a Ph.D. Candidate at University of Waterloo.
• 🔭 I'm broadly interested in deep learning and its applications to the field of computer vision.
• 🎯 I focus on the following areas, particularly in the context of autonomous robots:
  • 3D geometric understanding and perception
  • temporal motion forecasting
  • self-supervised representation learning

ssl-lanes's Issues

I can find data and utils folder in the code you have uploaded and hence its giving error for inference can you please provide the same

Inference gives NaN

Hi, thanks for your nice work.

When running python test.py -m lanegcn --weight=lane_masking.ckpt --split=val, the evaluation results are NaNs. Could you please tell me the cuda version of your experiment environment?

path/to/train

Thank you for your outstanding work

I would like to ask for the command mentioned in readme
horovodrun -np 4 -H localhost:4 python /path/to/train.py -m lanegcn

In this command,/path/to/train.py refers to which train.py in the warehouse, and I am not quite sure about the role of each folder. Thank you sincerely, if you can answer my question

visualization

Thank you for your excellent work.
I would like to ask where I can find “preds_base.p”，“gts_base.p”and so on.I would be very grateful if you could answer me.

Cannot load dataset

(/home/li/SSL-Lanes/.envSSl) [root@gpu14 final_baseline]# python preprocess_data.py -m lanegcn
lanegcn
Traceback (most recent call last):
  File "preprocess_data.py", line 415, in <module>
    main()
  File "preprocess_data.py", line 56, in main
    val(config)
  File "preprocess_data.py", line 118, in val
    dataset = Dataset(config["val_split"], config, train=False)
  File "/home/li/SSL-Lanes/baseline/final_baseline/data.py", line 29, in __init__
    self.am = ArgoverseMap()
  File "/home/li/SSL-Lanes/argoverse-api/argoverse/map_representation/map_api.py", line 71, in __init__
    self.city_rasterized_da_roi_dict = self.build_city_driveable_area_roi_index()
  File "/home/li/SSL-Lanes/argoverse-api/argoverse/map_representation/map_api.py", line 197, in build_city_driveable_area_roi_index
    city_rasterized_da_roi_dict[city_name]["da_mat"] = np.load(npy_fpath)
  File "/home/li/SSL-Lanes/.envSSl/lib/python3.7/site-packages/numpy/lib/npyio.py", line 440, in load
    pickle_kwargs=pickle_kwargs)
  File "/home/li/SSL-Lanes/.envSSl/lib/python3.7/site-packages/numpy/lib/format.py", line 771, in read_array
    array.shape = shape
ValueError: cannot reshape array of size 12910464 into shape (3043,4259)

Hi, I want to preprocess the data set, but run into this problem. Can you tell me how to solve it？
Thank you,

Training Process Terminated Unexpectedly with "Killed" Message

Hello,

I am attempting to replicate your code on my local machine, but I am encountering an issue during the training process. The error message is simply Killed, and the process stops. This is the full command and output:
(lanegcn) shanze@shanze-ROG-Zephyrus-G15-GA503QM-GA503QM:/media/shanze/UBUNTU2004/SSL-Lanes/ssl_pretext_tasks/lane_masking/final_task_masking$ /home/shanze/anaconda3/envs/lanegcn/bin/python /media/shanze/UBUNTU2004/SSL-Lanes/ssl_pretext_tasks/lane_masking/final_task_masking/train.py
Killed

For the data preprocessing step, I chose to directly "wget https://yun.sfo2.cdn.digitaloceanspaces.com/public/lanegcn/test_test.p" and put the .p file in the I put the .p file in the folder "/ssl_pretext_tasks/lane_masking/final_task_masking/dataset", and when I run train.py, the program should execute smoothly, but it always reports the error.

I am running on ubuntu 20.04 and have followed all the steps for configuring the libraries. I'd appreciate it if you could take a look at my question and reply, thank you very much!

Thank you in advance,
Shanze