archivedSep 1, 2025
Images Captioned to Minecraft
A deep learning project exploring how natural language descriptions map to Minecraft gameplay clips.
PythonTensorFlowHugging FaceDeep Learning
Highlights
- - Designed a data pipeline pairing gameplay clips with human-written captions.
- - Prototyped CNN and transformer-based models in Google Colab.
- - Planned retrieval and evaluation experiments for caption-to-clip matching.
Context
This project investigates multimodal learning by connecting visual game environments with natural language descriptions.
Current Work
Model architectures and training pipelines are still evolving. Results and a formal write-up are in progress.