Method for training a machine learning model to perform object detection

Broadly speaking, embodiments of the present techniques provide a method for unsupervised training of a machine learning, ML, model to perform object detection using unlabelled images. The method comprises method comprises obtaining a first training dataset comprising a plurality of unlabelled images, each unlabelled image containing at least one object and analysing the first training dataset by using an object detector module of the ML model to extract at least one bounding box for each unlabelled image and generate a pseudo-label for each extracted bounding box. A second training dataset is then formed using the unlabelled images of the first training dataset and their corresponding extracted bounding boxes and pseudo-labels and this is used to train the object detector module to output bounding boxes and pseudo-labels for input pseudo-labelled images.