The co-saliency detection technique is being applied in numerous applications, including image query, picture annotation, extraction of image/video foreground portion, surveillance, image retrieval, in order to discover most similar and salient patterns from a set of relevant image group. We proposed a method of co-saliency detection in an image, utilizing residual neural networks (ResNet 50 model) as classification model in this study. CLAHE (contrast limited adaptive histogram equalization) and Otsu segmentation technique are employed in this paper to better isolate and detect prominent object in image as pre-processing step. The result from these technique helps the model to understand the pattern more effectively. The fine tuning of the ResNet model has been done to produce more optimized and accurate result. The network was compiled with SGDM optimizer for more optimized result. Cosal2015 and MSRC dataset. are used to train and test the model for co-salient object detection and qualitative and quantitative assessment. The dataset used is divided into 70% and 30% for training and testing purpose respectively. The proposed model outperforms in terms of F1-score and MAE score with Cosal2015 dataset and MSRC dataset when compared with the other state of art methods. The proposed work is also efficient in terms of execution time and complexity.
The co-saliency detection technique is being applied in numerous applications, including image query, picture annotation, extraction of image/video foreground portion, surveillance, image retrieval, in order to discover most similar and salient patterns from a set of relevant image group. We proposed a method of co-saliency detection in an image, utilizing residual neural networks (ResNet 50 model) as classification model in this study. CLAHE (contrast limited adaptive histogram equalization) and Otsu segmentation technique are employed in this paper to better isolate and detect prominent object in image as preprocessing step. The result from these technique helps the model to understand the pattern more effectively. The fine tuning of the ResNet model has been done to produce more optimized and accurate result. The network was compiled with SGDM optimizer for more optimized result. Cosal2015 and MSRC dataset. are used to train and test the model for co-salient object detection and qualitative and quantitative assessment. The dataset used is divided into 70% and 30% for training and testing purpose respectively. The proposed model outperforms in terms of F1-score and MAE score with Cosal2015 dataset and MSRC dataset when compared with the other state of art methods. The proposed work is also efficient in terms of execution time and complexity.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.