Classify Retail Product Photos for an E-Commerce Marketplace
Overview
What this challenge is about.
Use a publicly-available product-image dataset (e.g., Fashion-MNIST extended, or a Kaggle e-commerce subset of around 10k images across 12 categories). Fine-tune a small pretrained CNN (ResNet18 or MobileNetV3) from torchvision. Evaluate top-1 and top-3 accuracy on a held-out 1,000-image test set. Compare results against the title-only baseline (provided as a CSV of category predictions). Deliver the trained model, a notebook, and a 3-page memo a non-engineer product manager can read.
The Brief
What you'll do, and what you'll demonstrate.
Train an image classifier that beats the title-only baseline on top-3 category accuracy by at least 10 points.
Earning criteria — what you'll demonstrate
- Fine-tune a pretrained CNN on a small classification task
- Apply standard image augmentation for robustness
- Evaluate classification with top-1 and top-3 accuracy + confusion matrix
- Communicate results to a non-technical product audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Computer Vision Engineer
Fine-tuning a CNN classifier with proper splits and a non-technical writeup is the entry-level CV engineering task every product team assigns to a new hire.
This challenge sharpens
- image-classification
- transfer-learning
- model-evaluation
Machine Learning Engineer
Train/val/test discipline and clean evaluation are habits MLEs carry into every project; this challenge builds them on a small, tractable problem.
This challenge sharpens
- pytorch
- model-evaluation
- data-augmentation
AI Engineer
Translating model results into a memo a PM can act on is the AI-engineer skill that gets work shipped.
This challenge sharpens
- image-classification
- model-evaluation
- transfer-learning