Abstract: An apparatus comprises a processing device configured to obtain a plurality of images each comprising at least a given portion of an instance of a computing device and to detect, utilizing an image classification machine learning model, at least a subset of the plurality of images having one or more designated failure characteristics. The processing device is also configured to create, utilizing a semantic segmentation machine learning model, a mask for each image in the subset of the plurality of images, the created masks characterizing locations of the designated failure characteristics in each image in the subset of the plurality of images. The processing device is further configured to generate an aggregated view of locations of the designated failure characteristics for the subset of the plurality of images utilizing the created masks, and to perform failure analysis for the computing device utilizing the generated aggregated view.