How it works
To create this machine learning system, we need to use ml5.js, a JavaScript library which provides access to machine learning algorithms and models in the browser, as well as MobileNets. Image classification is a machine learning problem where we define a set of targets, then train a model to recognize them using sample images. In our case, targets are cities : we need to establish a large dataset containing multiple images of different cities and let the AI learn from that data. Indeed, each city has a pattern based on its specific characteristics, such as monuments, architecture, famous places and so on. The more images the AI sees, the better its gets in its prediction.
To summarize :
1. We give the AI a broad image database with various cities and different places for each city.
2. The AI analyzes each image and integrates the patterns of each city to know how they look like.
3. The AI can recognize a city by comparing the image which has been uploaded to similar images in its database.
| City |
Images |
| Paris |
|
| New York City |
|
What about coding ? We can use ml5.js by referencing it in our project, using a script tag in an HTML file as below :
To go further, here are some helpful videos to understand better image classification using artificial intelligence :
