How we managed to create this AI product in three weeks:

June 15 - June 17

Drafting Project Ideas

The topic of our AI Camp was computer vision, so our project had to correspond to this topic. Given this, there were many topics we could choose from. Before deciding on identifying whether or not someone was wearing a mask, we spent a few days brainstorming on other possible topics and considered detecting animals and identifying poisonous mushrooms.

June 17 - June 19

Collecting/Cleaning Data

We then collected and cleaned data. To be more productive, we split into two groups: one group worked on collecting images of faces with masks, and the other group worked on collecting images of faces without masks. The two groups also collected data in different ways. The former used a Chrome extension to download images into a zip file, and the latter group extracted data by using Python to request a Bing API. Subsequently, both groups had to sift through their data and delete unnecessary data.  

June 19 - June 22

Labeling Data

We used Labelbox to label our photos of faces with masks versus without masks.

June 22 - June 25 

Training Data 

We then prepped for the training by downloading the labeled images, converting them into the YOLO format, configuring YOLO, and installing darknet. After that, Rahul and Richard trained our model. 

June 26

Testing Data

We then created tested our product. After testing our product, Madeleine constructed a confusion matrix to analyze our data. 

June 29 - July 3

Deploying Model

After creating our model, we then deployed it onto this website for the public to use. We were also thinking of deploying our model to a small store for use.