search

argoverse / av2-api

Argoverse 2: Next generation datasets for self-driving perception and forecasting.
https://argoverse.github.io/user-guide/
MIT License
307 stars 71 forks source link

Using the rust/ pytorch dataloader #182

Closed SM1991CODES closed 1 year ago

SM1991CODES commented 1 year ago

Hi, My datasets are at : image

Then I try this: `logging.basicConfig(level=logging.INFO) logger = logging.getLogger(name) logger.info("Starting detection data-loader example ...") root_dir = Path("/home/data/argoverse/av2/part1") data_loader = DetectionDataLoader(root_dir, dataset_name="sensor", split_name="train",) for i, sweep in enumerate(tqdm(data_loader)):

4x4 matrix representing the SE(3) transformation to city from ego-vehicle coordinates.

    city_SE3_ego_mat4 = sweep.city_SE3_ego.matrix()

`

But I get an error: image

Please let me know what is wrong here. I checked the tutorial and it says: root + {dataset_name} + {split_name} should end up inside one of the part folders train/ val directory. I think I am doing just that.

Best Regards Sambit

SM1991CODES commented 1 year ago

The code did not get formatted properly: image

SM1991CODES commented 1 year ago

I got it. I did import tqdm instead of from tqtm import tqdm. Now it works.

benjaminrwilson commented 1 year ago

Glad you got it figured out!

SM1991CODES commented 1 year ago

Please take a look at the other issue I face now regarding using the old convenience functions with the new Pytorch data loader. Kind of stuck and following an ugly approach right now - create both DetectionLoader and SensorLoader objects.