I trained a neural network to predict locations in Counter-Strike using over 1 million images.

👉🏼 My previous CS 2 AI map prediction:    • I made an AI to predict Counter Strik...  


0:00 Introduction
0:45 Capturing Counter-Strike 2 Screenshots
2:35 Reviewing the Dataset
3:42 Normalizing the data
5:13 Model setup
6:53 Training Run
8:31 Reviewing the Training Results
10:07 Result Visualization with UMAP
12:42 Subscriber Battle, Outro, Recommendation (not sponsored)


View my UMAP embedding visualization (only for desktop) - not even the homepage works on mobile and I’m on vacation and can’t fix it - https://cstrike.win/
Map Location Dataset - https://github.com/Programming-with-J...
Subscriber battle -    / @tylermommsen  

Music: https://archive.org/details/Home-Odyssey
HOME - "Intro"
HOME - "Oort Cloud"
HOME - "Tides"
HOME - "New Machines"
HOME - "Half Moon"
HOME - "Resonance"

Credits:
Source 2 Viewer - https://github.com/ValveResourceForma...
GeoGuesser AI -    • AI Learns to Play GeoGuessr  
ResNet vs. EfficientNet -   / image-classification-resnet-vs-efficientne...  
UMAP Explanation Video -    • Latent Space Visualisation: PCA, t-SN...  
Weights and Biases - https://wandb.ai/
Weights and Biases Tutorial -    • Log Your First Run With W&B  

Support me so I can quit my job and go live in the woods:
patreon.com/ProgrammingWithJulius