monocular-vo project VAMR 2024

Demonstration

A video demonstration of the pipeline in action can be found below (click here if the embed doesn't work)

A breakdown of the code and features incorporated can be found in the project_report

Specifications of the computer from the demo-video

• Ryzen 97950x16/32 cores/threads at 5.2GHz
• The threads of the python threading tool may span over multiple cores, so the exact number of CPU threads is not known, but overall CPU usage was about 30%
• 64GB of 6000MHz dual channel memory, only about 1Gb of RAM usage, however

Dataset setup

Create a folder called datasets, download and unzip the datasets found here into that folder:

Description	Link (size)
Parking garage dataset (easy)	parking.zip (208.3 MB)
KITTI 05 dataset (hard)	kitti05.zip (1.4 GB)
Malaga 07 dataset (hard)	malaga-urban-dataset-extract-07.zip (2.4 GB)
Own Dataset	https://share.easywin.ch/s/tXC9KEZ0 (1.2 GB)

This should get you folders datasets/kitti, datasets/malaga-urban-dataset-extract-07, and datasets/parking for use in the pipeline. The additional dataset recorded for this project is in a folder own_dataset with subfolders own_dataset/ds1, own_dataset/ds2, and own_dataset/calibration.

Running the script

1. Setup environment

   A conda environment is provided for installing the dependencies required.

   conda env create --name vamr --file=vamr-env.yml
   conda activate vamr
   

2. Run VO pipeline

   Run main.py.

   python3 main.py
   

   You are prompted to insert a number determining what dataset should be evaluated. Before VO begins, the raw images are loaded into memory.

3. Run Evaluation

   Run evalution.py.

   python3 evaluation.py
   

   You will be again prompted to insert a number determining what dataset should be evaluated. Note that you must have run the pipeline first using main.py or this will not work.