I am interested in machine learning. I’ve finished most of the Udacity “Intro to Machine Learning Course”. I’ve been thinking of ways to get my feet wet in machine learning. A practical project that I can start and finish that will give me some hands on experience.
Hello TrafficFlow (Traffic + TensorFlow)
I’ve built an Android app, Traffcams, that lets people view traffic images from traffic cameras. I’ve done the TensorFlow tutorial walking through image recognition. So I’m thinking that I can modify that tutorial to tell me if an image from a traffic camera contains a lot of traffic. My first step in training a TensorFlow model is collecting the data. I wrote a Python script that simply saves an image to disk from a given URL.
I have this script cron’d on a Ubuntu server. It runs every 4 minutes saving an image from this camera, which means I’ll save 360 images per day. I’ll probably throw away the night pictures (sunset to sunrise is about 8 hours)…so I’ll acquire about 240 usable pictures per day. I’m predicting I’ll need about 2,000 to 3,000 images to train a model. I’ll play it safe and say I’ll need 3,000 images. In 12 and a half days, I’ll have enough data to train.
My next step is to manually classify these images as having a lot of traffic (1) or not (0). Sounds monotonous.