-
The Concept Art of Astra Tour
-
Creating the Model of our ririn.
-
First Prototype of the Mapping System
Inspiration
We were inspired by the task itself. The idea is to embed the taking into the photos into the game narrative. Also some ideas were borrowed by the
What it does
Prepare to enter the shoes of the ririns - a peaceful race of aliens that seek to explore and document Earth. Using their smartphones the players have to go to specific event locations were they have to complete given challenges - for example puzzles. The player has to go to specific locations, which are marked on a map and take photos of certain problematic locations, which are then processed by a neural network to fix certain mislabeled locations on OpenStreetMap.
How we built it
Work was split into the game implementation and the neural network implementation between the two team members. The game part was done via Blender, Mapbox and Unity. The neural network was implemented from scratch using Golang and the provided datasets.
Challenges we ran into
Creating and training neural network, and finishing within the allocated time frame. Creating a game concept that fits with the challenge. Splitting the workload efficiently.
Accomplishments that we're proud of
Surviving 36 hours in StudiTUM. Training a ML model with very limited priror knowledge.
What we learned
Neural networks are not as complicated as initially thought, but one mistake costs a lot.
Log in or sign up for Devpost to join the conversation.