Google's open-source TensorFlow project has an exceedingly wide range of potential use cases, and startup Connectera has seized upon a particularly unique one; dairy farming. Connectera has created a special sensor that links to a mobile app, called Ida, and analyzes cows' movements and activities to give farmers insights on when cows may need special care, when cows are sick, and what kind of special care will be the best for helping the cows get back to maximum productivity as quickly as possible. The company was started in the Netherlands as a joint project between Yasir Khokhar and Saad Ansari. Yasir got the idea from living on a farm and noticing that a cow's health can make a drastic difference in how much time a farmer has to invest in that cow, and how much milk it can produce.
To help publicize Connecterra and its Ida product, Google allowed Yasir Khokhar and Saad Ansari to author a blog post, and made a video about one satisfied user of the Ida app, dairy farmer Neils Rutten. The video shows off the special sensor that goes around cows' necks and connects to Ida. Rather than giving Ida users the movement data gathered by the sensors in raw format, Ida uses TensorFlow as a backend to train an individual machine learning model for each cow that figures out what a cow is doing at any given time, and analyzes activity patterns to predict susceptibility to disease, when a cow may be sick, and other complications that could affect a cow's productivity and care. It can also tell how much a cow is eating and drinking, and how fertile it may be at any given time. These are all variables that can affect the quality and quantity of milk that a cow can give.
TensorFlow is one of the most flexible AI templates in the industry, and can even be used on a user or company's own hardware, or on Google's. The company created special system-on-chip units for TensorFlow usage, called Tensor Processing Units. TensorFlow is fully open-source and available for free, so unique and niche use cases like this will likely continue to emerge well into the foreseeable future.