The study of jet-inflated X-ray cavities provides a powerful insight into the energetics of atmospheres of early-type galaxies and the AGN feedback phenomenon. Properly estimating their total extent is, however, non-trivial, prone to biases and nearly impossible for poor-quality data. For these reasons, we have decided to harness the power of machine learning to tackle this problem. Using artificially generated images, we have trained a convolutional neural network to produce pixel-wise predictions capturing both the position and extent of detected X-ray cavities. Furthermore, we present how the network performs on real Chandra images of early-type galaxies.