- Projects 22
- Followers 72
DesertShooterXR
First Person Shooter game for Windows 10. Game uses Tensorflow Object Detection to control DesertShooter Agent. ...learn more
Project status: Under Development
Game Development, Artificial Intelligence
Groups
Early Innovation for GameDev
Intel Technologies
Other,
OpenVINO
Overview / Usage
The beginning scene that consists of a military base in a desert. A star-ship lands on the helipad in front of you when the game starts. Enemy robots spawn from that spot every few seconds and they will invade your position and track you. Use DesertShooterXR tensorflow based controller to fire a bullet to destroy the robots. The cross-hair shows where you aim, through the cross-hair is deliberately rendered close to you to force you to shift your gaze focus between the cross-hair and objects in the distance, similar to a real gun trigger. DesertShooterXR also includes spatial sounds distributed across the scene to provide a better feeling of immersion (i.e. wind sounds in the tower and canyons, machine sound in the factory, pressure sounds in the pump hacks)
The DesertShooterXR tensorflow based controller is designed to handle the following actions in the game:-
- Aiming the gun - in order to look around in the game we are using object detection
- Moving the player - to instruct the player to move forward in the game, we are using detection of the index finger
- Shooting the gun - using the mouth open gesture to control shooting of the gun.
Methodology / Approach
This project runs on Windows 10 PC's, it is buit using using Unity_2018.3.9.f1 and the Windows 10 SDK. DesertShooterXR uses The model used for object detection here is MobileNet in combination with Single-Shot Multi-Box Detector (SSD) for image localization. Simple cloud system and spatial sounds are also integrated into this project.
Technologies Used
- Windows 10 Visual Studios 2017
- Windows 10 SDK
- Unity_2018.3.9f1
- Tensoflow
- Intel OpenVINO Toolkit
- Intel i7 8th Gen Intel HD Graphics 620
Other links
Collaborators