VMBA

The ultimate goal of this project is to deduce customer behaviors during their shopping trips using sequences of images acquired from cameras mounted on shopping carts. This information can be used to perform Market Basket Analysis, helping retailers improve space management and marketing strategies.

Demo

The demo available in this repository was built using Open Neural Network eXchange (ONNX), which allowed us to export models via Python in ONNX format and subsequently use them on a web page.

On the page, you can select an image from the dataset via a dropdown menu. This image is passed to AlexNet and ResNet-50, the models that achieved the best results. The screen displays the output of each model, showing prediction percentages for each class and the actual label of the image. Additionally, if the model correctly classifies the input, the box will turn green; otherwise, it will be red.

Another feature allows users to upload their own images to be classified by the two models.

How to use

1. Run npm install to initialize the repository and download all dependencies.
2. Run npm run build to build the project and prepare the web page.
3. Run npm run serve to start the server and make the program available in the browser.
4. Visit localhost:8080 in your browser to view the demo.