stasiekkolodz / dml-multitasking-project

Multi-Task Image Colorization and Classification

This work aims to examine the effects of applying multi-task learning to the task of image colorization and classification. Specifically, we train and compare the performance of a model trained to perform both tasks with models trained to perform each task individually. We then investigate whether the multi task model becomes better at any of the two tasks than the corresponding single-task model.