DETECTION.md

Detection

This file provides instructions on how to perform detections upon BOP19 Dataset

Setup

1. Set up mmdetection baseline according to here.

2. Download all testsets in detection/bop19 from here(or you could download somewhere else and set up a soft link).
   Take lm as an example

   export SRC=http://ptak.felk.cvut.cz/6DB/public/bop_datasets
   wget $SRC/lm_base.zip        # Base archive with dataset info, camera parameters, etc.
   wget $SRC/lm_models.zip      # 3D object models.
   wget $SRC/lm_test_all.zip    # All test images ("_bop19" for a subset used in the BOP Challenge 2019).
   wget $SRC/lm_train.zip       # Training images.
   
   unzip lm_base.zip            # Contains folder "lm".
   unzip lm_models.zip -d lm    # Unpacks to "lm".
   unzip lm_test_all.zip -d lm  # Unpacks to "lm".
   unzip lm_train.zip -d lm     # Unpacks to "lm".
   

   Note: All datasets should be in detection/bop19 whose file system should be as following:

   ├── lmo
   │   ├── test
   │   │   ├── <scene_id>
   │   │   │   ├── rgb
   │   │   │   │   ├── <img_id>.png
   │   │   │   │   └── ...
   │   │   │   ├── mask_visib
   │   │   │   ├── depth
   │   │   │   └── ...
   ├── ycbv
   │   ├── test
   │   │   └── ...
   └── ...
   

3. Find the root directory of your mmdetction path

   # Please specify the path below
   export MMDETECTION_PATH=<your mmdetection path>
   export DETECTION_PATH=<detection>
   
   # Set up soft link here to self-define the bop19-dataset
   cd ${MMDETECTION_PATH}/mmdet/
   ln -s ${DETECTION_PATH}/datasets
   
   # return to your work place
   cd ${DETECTION_PATH}
   

4. Download models from here (Password: 7yjb) and store them in ${DETECTION_PATH}/models

5. Generate the detection results by

   pip install tqdm -y
   python inference.py <dataset name>
   

6. Then you have your detection results in ${DETECTION_PATH}/out, and the file system is as below:

   ├── hb
   │   ├── hb_<scene_id>_detection_result.json
   │   └── ...
   └── ...
   

   You can go on test on the Pose Part using these files.