How to build a neural network to detect an object in a photo?

Question

Pattern recognition

I want to train a model using the COCO 2017 dataset. The point is that there are several objects in all the photos, and this is what the architecture of the neural network should look like in order to correctly predict?

Answer 1

Take for example the RetinaNet model

RetinaNet architecture picture

Ready laptop in Google Colab: RetinaNet + COCO 2017

Answer 2

Okay, google, convolutional networks .