If you are looking for the old repository, pre March 2018, it is here
This project contains Python and MATLAB scripts for processing the Active Vision Dataset
A starting point for loading our data in python for detection
or classification of instances. See data_loading/demo.py
With this code, you can:
- Get an iterable data structure that gives images and bounding box labels
- Use the data structure to crop boxes for classification
- Apply pre-defined and custom transforms to the images and labels
Use the code here to visualize our data. See the README in this folder for examples.
To get started, edit the ROHIT_BASE_PATH variable in run.py and/or init_paths.m
to point to the directory that contains our data. i.e. the parent directory
that the directory for each scan.
ROHIT_BASE_PATH = /path/to/dataset/
Everything is a function. See examples below for how to run certain functions, or look at the README in the visualizations directory
To run a python function from the command line:
python run.py function_name scene_name
- Start MATLAB
- Run
init_paths - Call the desired function