How to install rpl_vision_utils package?

[GDJUpUp]

How to install rpl_vision_utils package?`

Environment

• OS: Linux (Ubuntu)
• Python: 3.6.15
• PyTorch: 1.10.1

Issue Description

I am trying to reproduce your work and run the evaluation script eval_collector.py. After successfully setting up the environment and installing dependencies, I encountered the following error:

Traceback (most recent call last):
  File "deoxys-lang4sim2real/deoxys/scripts/eval_collector.py", line 14, in <module>
    from deoxys.utils.data_collection_utils import (
  File ".../deoxys/utils/data_collection_utils.py", line 16, in <module>
    from deoxys.utils.obj_detector import ObjectDetectorDL
  File ".../deoxys/utils/obj_detector.py", line 3, in <module>
    from yolov5.utils.general import non_max_suppression
ModuleNotFoundError: No module named 'yolov5'

I resolved the yolov5 issue by running:

pip install yolov5==6.0.1 --no-deps

However, I then encountered another missing dependency:

  File ".../deoxys/envs/franka_env.py", line 99, in import_realrobot_pkgs
    from rpl_vision_utils.networking.camera_redis_interface import (
ModuleNotFoundError: No module named 'rpl_vision_utils'

Questions

1. Where can I find the rpl_vision_utils package? It doesn't seem to be available on PyPI or any public GitHub repository.

2. Is it possible to run the evaluation in simulation-only mode without requiring real robot hardware access?

3. Could you add the missing dependencies (rpl_vision_utils) to the installation instructions in README?

Command Used

python deoxys-lang4sim2real/deoxys/scripts/eval_collector.py \
    --ckpt LOCAL_LOG_DIR/26-01-07-BC-image-frka-pp/.../itr_150.pt \
    --obj-id 1 --env frka_pp --state-mode 1 \
    --task-embedding lang --lang-prefix "Real:" \
    --max-path-len 20 --num-tasks 2 \
    --eval-task-idxs 0-0 --num-rollouts-per-task 10 --gpu 0

Thank you for your great work and for making the code available!