Enhancing Semantic Segmentation through Reinforced Active Learning: Combating Dataset Imbalances and Bolstering Annotation Efficiency

Abstract

This research addresses the challenges of training large semantic segmentation models for image analysis,focusing on expediting the annotation process and mitigating imbalanced datasets. In the context of imbalanceddatasets, biases related to age and gender in clinical contexts and skewed representation in natural images can affectmodel performance. Strategies to mitigate these biases are explored to enhance efficiency and accuracy in semanticsegmentation analysis. An in-depth exploration of various reinforced active learning methodologies for imagesegmentation is conducted, optimizing precision and efficiency across diverse domains. The proposed frameworkintegrates Dueling Deep Q-Networks (DQN), Prioritized Experience Replay, Noisy Networks, and EmphasizingRecent Experience. Extensive experimentation and evaluation of diverse datasets reveal both improvements andlimitations associated with various approaches in terms of overall accuracy and efficiency. This research contributesto the expansion of reinforced active learning methodologies for image segmentation, paving the way for moresophisticated and precise segmentation algorithms across diverse domains. The findings emphasize the need fora careful balance between exploration and exploitation strategies in reinforcement learning for effective imagesegmentation.