Closed Sar-thak-3 closed 1 month ago
Please take a look at this document for more details about BOP format. You can also use this script to extract *.zip after downloading from HuggingFace get the dataset in BOP format.
For testing the toolkit, you can play with the MegaPose's predictions available in the toolkit. For example:
python scripts/eval_bop19_pose.py --renderer_type=vispy --results_path ./bop_toolkit_lib/tests/data/ --eval_path ./bop_toolkit_lib/tests/data/ --result_filenames cnos-fastsammegapose_lmo-test_16ab01bd-f020-4194-9750-d42fc7f875d2.csv --num_worker 10
Feel free to re-open the issue if the error persists.
Description: I'm currently grappling with an issue concerning the directory structure necessary for evaluating my pose estimation model using the BOP (Benchmark for 6D Object Pose Estimation) toolkit. It seems that the organization of my files isn't aligning properly with the toolkit's expectations, leading to errors and hindering accurate evaluations.
my
config.py
fileI only wanted to evaluate the model on ycbv testing data, which is available at provided BOP website https://huggingface.co/datasets/bop-benchmark/datasets/tree/main/ycbv
And the
/predictions
folder contains single.csv
file for predicted output by the model.Also one more thing
python scripts/eval_bop19.py --renderer_type=vispy --result_filenames=NAME_OF_CSV_WITH_RESULTS
This command runseval_bop19.py
, instead there should beeval_bop19_pose.py