Transfer learning for endoscopy disease detection and segmentation with mask-RCNN benchmark architecture

We proposed and implemented a disease detection and semantic segmentation pipeline using a modified mask-RCNN infrastructure model on the EDD2020 dataset. On the images provided for the phase-I test dataset, for 'BE', we achieved an average precision of 51.14%, for 'HGD' and 'polyp' it is 50%. However, the detection score for 'suspicious' and 'cancer' were low. For phase-I, we achieved a dice coefficient of 0.4562 and an F2 score of 0.4508. We noticed the missed and mis-classification was due to the imbalance between classes. Hence, we applied a selective and balanced augmentation stage in our architecture to provide more accurate detection and segmentation. We observed an increase in detection score to 0.29 on phase -II images after balancing the dataset from our phase-I detection score of 0.24. We achieved an improved semantic segmentation score of 0.62 from our phase-I score of 0.52.