https://github.com/lorigreensmith06/AIML_Capstone
For my final project in the AI & Machine Learning certification program, I created a prototype for an e-commerce platform called BuyIt! — a tool designed to help customers find vendors at local markets, like farmers’ markets and small stores. I chose this problem because of my strong interest in small businesses and my desire to help people connect locally.
I also wanted to explore how AI can be used to classify images, and this project gave me a chance to dive into computer vision for the first time. I’m proud that I gained a solid understanding of how AI works — and even more excited that it was so effective in solving the problem.
Algorithms I Tested
I implemented and compared the following classification algorithms to determine which best classified product images:
- Decision Trees
- Logistic Regression
- K-Nearest Neighbors (KNN)
- Support Vector Classifier (SVC)
- Neural Networks
- Transfer Learning (fine-tuning a pre-trained MobileNetV2 model)
Despite some challenges with the dataset — including noisy labels and uneven class distributions — I was able to achieve 87% accuracy using transfer learning. By fine-tuning MobileNetV2 on my custom image set, I was able to significantly boost performance while reducing training time.
I further improved accuracy to 92% using ensemble methods, combining several of the simpler models that performed well in specific categories.
Reflection and Future Work
While all of this might not seem directly related to my previous work, I was genuinely curious about AI and wanted to explore it more deeply — because I’m certain it will play a major role in our future. I hope that I may have a place in that future too, especially by helping people find ways to insert their creativity into AI.
I’m especially interested in the overlap between computer graphics and neural networks, and I feel like I’ve only just scratched the surface. This project was a starting point, and I’m excited to keep going.
