NVlabs / VoxFormer

Official PyTorch implementation of VoxFormer [CVPR 2023 Highlight]
Other
1.07k stars 87 forks source link

evaluate problem #27

Open a2801603396 opened 1 year ago

a2801603396 commented 1 year ago

When I run test.py, I will report an error: TypeError: string indices must be integers In the following code:

    for result in results:
        !!self.metrics.add_batch(result['y_pred'], result['y_true'])!!
    metric_prefix = f'{result_name}_SemanticKITTI'

In semantic_kitti_dataset_stage2.py and semantic_kitti_dataset_stage1.py. The function is “evaluate”.

Whether it's me testing qpn or voxformer-S, this error occurred in the same location.

a2801603396 commented 1 year ago

The first result is a string with the content 'y pred’。 Results is a list that alternates between 'y pred' and 'y_true'.

a2801603396 commented 1 year ago

How much memory did you use during the testing?I ran out of 16GB of memory and the verification set was full of memory.The test set is even more difficult to run.

RoboticsYimingLi commented 1 year ago

Please follow our tutorial exactly https://github.com/NVlabs/VoxFormer/blob/main/docs/getting_started.md. Thank you!

RoboticsYimingLi commented 1 year ago

Thank you for your interest in our paper. Please make sure to follow our tutorial exactly and use the dist_test.sh script to obtain evaluation results.

Made-Gpt commented 10 months ago

When I run test.py, I will report an error: TypeError: string indices must be integers In the following code:

    for result in results:
        !!self.metrics.add_batch(result['y_pred'], result['y_true'])!!
    metric_prefix = f'{result_name}_SemanticKITTI'

In semantic_kitti_dataset_stage2.py and semantic_kitti_dataset_stage1.py. The function is “evaluate”.

Whether it's me testing qpn or voxformer-S, this error occurred in the same location.

I also have this problem, it occurs when I was trying to run test with one GPU, I change the dist_test.sh: `CONFIG=$1 CHECKPOINT=$2 PORT=${PORT:-29503}

PYTHONPATH="$(dirname $0)/..":$PYTHONPATH \ python -m torch.distributed.launch --master_port=$PORT \ $(dirname "$0")/test.py $CONFIG $CHECKPOINT --eval bbox`