[Project] [Paper] [arXiv]

This is the official code of our APIN 2022 paper "Dense Depth Estimation from Multiple 360-degree Images Using Virtual Depth".

• Ubuntu 18.04
• C++11 Compiler
• OpenCV > 3.0 (Tested with OpenCV 3.4.6.)

Clone the repository:

git clone https://github.com/ysy9997/360Depth.git

You can simply execute build.sh to build this program.

cd 360Depth
chmod +x build.sh
./build.sh  

This will create libDEMO.so at lib folder and two executable image in current folder.

You can simply run like below.

./image [the number of cameras] [max depth] [rigid] [save folder path] [0-th images folder] [1-th images folder] ...

If set 'rigid' as 1, it means rotation matrix is identity, and translation vector is [0, 0, -1]. If set 'rigid' as 0, this application will estimate rotation matrix and translation vector.

The max depth needs to be divided as a baseline between 0-th and 1-th cameras. (e.g. When the baseline is 0.5m and the furthest distance is 10m, then, the max depth should be 20.)