At a 48-hour hackathon, our team built a bottle-recognition prototype. I prepared the image-processing pipeline and worked on fine-tuning its classifier.
My role: Data pipeline and model tuning · Team hackathon project
Highlights
48-hour hackathon winning project
AI-driven bottle classification for retail return flow
My contribution
Prepared image data for model training and testing.
Fine-tuned and evaluated the bottle-classification model.
Worked with the team to integrate the classifier into the demonstration.
Technical context
The prototype used image classification to recognize bottles for a potential return-system workflow.
Different bottle shapes, labels and image conditions made data preparation and model behavior important parts of the implementation.
Our team won an award for the project at the hackathon.
