Clothing Finder

Inspiration

We were inspired by the vision of e-commerece that InditexTech put forward at the opening ceremony.

What it does

The project detects if there are duplicate/similar images of a clothing garment and groups them together.

How we built it

We built the model using Python. We used a convolutional neural network (CNN) as well as supervised learning classifier, K nearest neighbour (KNN). We also developed a website where we display necessaary information regarding the ML model.

Challenges we ran into

We initially ran into problems with figuring out what to train our data off of.. Additionally, the timing was difficult for us as every error we dealt with in the code, meant that we had to retrain our model. Training our model was extremely time comsuming so a lot of time was wasted.

Accomplishments that we're proud of

We are proud that our ML model is able to process images and is able to group images together based off of common colours and shapes.

What we learned

We have learned what was a CNN and how to use one, as well broadened our horizons with the different machine learning methods when we were researching for a solution.

What's next for Clothing Finder

The next step for Clothing Finder is to further iterate on the design. Although the model is able to detect and recognize similar colours, patterns, and shapes, it isn't completely accurate. After the hackathon, we plan on further working on the design, to iron out any creases within the model so we can improve its accuracy.

Built With

• numpy
• pandas
• python
• requests
• scikit-learn
• tensorflow
• vscode