Yesterday, we had our group’s meeting with our Python Machine Learning Labs teacher to update him with the progress of our project. We are building a Face Detection Application that we could either deploy on the web or as a mobile application. For my part, I used my Data Science (DS) classmate’s model that was pre-trained on SSD MobileNet V2 FPNLite 640×640 with 50000 steps (~0.7s per step) for my web app deployment.
Here is a snippet of the meeting’s recording, showing the part where I presented my work on react.js web app that I have deployed in Heroku using Github workflow actions, something we recently learned this week on our DevOps class.
The teacher was not happy with react.js app due to the overhead of the browser. Heroku was also not the best place to deploy the web app. He was also a bit confused about my colorful bounding boxes coz I was randomizing a list of color codes to display it. He thought it was some error with the detection. LOL! He suggested instead that we go with AWS technologies, Lambda functions in particular, and use Python instead. I think that this was a good exercise for me, and the group. Even though, we are not going to use this deployment in our project, it was interesting to use what I have learned this week in our DevOps class on the subject of CI/CD, and using Github workflow actions. 🙂
Here is the link to the Heroku deployment, if you want to check it out: https://diversity-face-detection.herokuapp.com/
Note that the app takes some time to load, and you will have to allow access to your camera. Also, sometimes, you might need to reload the page to see the bounding boxes.