Not only large companies like OpenAI can create AI models. You too can experiment with machine learning at home and create small, useful applications. With Google’s Teachable Machine service, this is possible free of charge and without much programming knowledge.
In this article, we show how to create a model that recognizes objects in images. Specifically, it is about distinguishing between dogs, cats, lions and birds. The AI should also classify bird species as birds that do not appear in the training data set. We also explain how to collect enough footage for training. At least 200 photos are usually required for each object to be recognized; the more templates, the better the results.
The training can be carried out on a normal PC. Mini computers like the Raspberry Pi are too weak for that. But we also show how, after training on a mediocre PC, the finished model can be exported in Tensorflow Lite format and run on the Raspberry Pi. Any photos of animals can then be photographed with a Raspi camera and the AI tries to classify them. The principle can be transferred to other areas such as audio or gesture recognition. With enough training images, face recognition would also be conceivable to search through the mountains of images at home for people.
