| Popis: |
In the era of artificial intelligence, industries are looking for the high performance of semantic segmentation methods to narrow down the complexities of real-world imaging and computer vision applications. This analysis became a state-of-the-art choice for growing industries to model in the automobile, medical and agriculture sectors etc. Semantic segmentation is the field of image analysis which was implemented using traditional strategies. Now the industries were taken a turn to implement learning algorithms like machine learning, deep learning, etc. From the past decade onwards, many researchers have worked on semantic segmentation using Deep learning strategies i.e. supervised, unsupervised, semi-supervised, or weakly supervised learning. The review is needed to discuss existing strategies and their limitations to improve the real-world applications performance. This paper categorized the learning strategies according to data availability of the problem. It also discussed the procedures and limitations of the existing work. It concludes the future directions of the research on semantic segmentation. |
