Q-SCAN

Our quantum module, called Q-SCAN, is a quantum image segmentation module designed to efficently detect flooded areas from satellite images.

Image segmentation is a computer vision technique that divides an image into several regions or segments to simplify its representation and facilitate analysis.

Imagine you want to automatically detect objects in images (forest fires in aerial photos, floods in satellite images, defects in photos of manufactured parts). To do this, segmentation algorithms divide the image into different groups of pixels called segments. Each segment corresponds to a part of the image with common characteristics, such as color, texture or intensity. This makes it possible to automatically identify and delimit objects in the image, facilitating image analysis.

Classical segmentation methods are widely used in certain contexts, but they have their limits.
On the one hand, traditional state-of-the-art methods, offering the best performance in terms of quality and accuracy of results, are based on supervised learning models. As a result, they are very costly to implement, due to the need for manually labeled data and the high cost in human and material resources required to train them before they can be deployed and used. On the other hand, existing classical unsupervised methods for segmenting images do away with the problems of labeled data and training, at the cost of the quality of the results. Indeed, the latter still struggle to deliver good performance, not least because of their sensitivity to noise and variations in brightness.

Our Q-SCAN quantum module uses an innovative unsupervised quantum segmentation method, offering an exceptional quality/cost compromise.
Q-SCAN is based on a graph problem that treats the spectral and spatial information contained in the image on an equal footing, thus improving the quality and robustness of the results obtained. This innovative formulation is ideally suited to quantum computing, enabling the problem to be solved efficiently without consuming excessive computing capacity. Q-SCAN therefore offers the best of both worlds: a low-cost, easy-to-implement, unsupervised solution with a quality comparable to that of state-of-the-art supervised methods. What's more, the architecture of this solution makes it versatile and flexible, enabling it to easily address a wide range of use cases without any reformulation effort.