MultiChexNet is a model that is able to classify, detect, and segment chest related diseases in one forward propagation. We apply the concept of the MultiNet architecture to medical images. This architecture did not only speed up inference time for the three tasks but we also saw improvements over training individual models at a time.
Segmentation results Detection results
Training steps
Multiple dataset were used for the training process, and they are the following: