Some time ago, we have created a blog post about a selfie camera for Chicken (Chickfie), maybe you have wondered what the purposes of these cameras are, besides taking photos of chicken for fun. Now, one year later, we have the answer for you – for Machine Learning.
Basically we tried to train the machine (computer) to recognize presences of chicken by using a lot of photos taken by the „Chickfie“ camera. We are delightedly to find out how accurate the results are.
Below is a short description how we did it.
First of all, we have to collect a lot of „chickfies“
We have used around 180 images.
Afterwards we teach the machine where the chicken are by marking them one by one.
so that machine knows how the animal looks like and what to look for in a picture.
The computer processes these images and start to learn
We have used an object detection algorithm to train the model (mathematical one, unrelated to fashion).
Now it’s your turn to try
How many chicken are there? (Move the slider to see the answer)
So sometimes the machine can be smarter.
We think that machine learning or artificial intelligence technologies are useful for farmers and animal lovers.
We are very excited about the potential use cases, for example, we can now tell how many chicken are in the coop and how many are still playing outside, whether the coop door is closed after dark, or even answer the impossible question „who laid the biggest egg of the day?“