FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4207029713> ?p ?o ?g. }

• W4207029713 abstract "Pterygium is an eye condition in which abnormal tissues grow from the medial canthus region that gradually spread towards the pupil region as the disease becomes more severe. Generally, it is a harmless condition but if the abnormal tissues have encroached onto the pupil region, the patient's vision will be severely affected. It is important to diagnose this condition at the early stage so that remedial steps can be administered to curb the growth of the tissues. Therefore, it is crucial to be able to measure the severity level of this condition accurately, which can be inferred from the size of the infected area. However, manual segmentation of the region of interest is a tedious task and usually results in different segmentation opinions between the health practitioners. Hence, an automated approach using deep convolutional neural networks is proposed to extract the infected region. This work takes a high-resolution approach, in which the deep model utilizes a set of multi-scale feature maps that still retain the highest possible resolution until the last convolutional layers. It is in contrast to the standard approach, whereby the feature map size will be reduced to a compact representation before it is being scale-up to a higher resolution. Furthermore, a set of group and shuffle operations has been embedded into the network to improve model capability in learning diversified feature maps. The networks will have a reduced dependency on a set of global features, whereby it is forced to extract meaningful features from a few separated groups before they are combined through a concatenation operator. The results show that a base HRNet-32 network with four sets of group convolution followed by a shuffle operation returns the best performance in terms of Accuracy, Mean IoU and Jaccard Index with performance values of 0.9369, 0.8707 and 0.8088, respectively. For future work, the optimal placement of the group and shuffle operations can be fine-tuned to increase the segmentation accuracy." @default.
• W4207029713 created "2022-01-26" @default.
• W4207029713 creator A5050837249 @default.
• W4207029713 creator A5054106321 @default.
• W4207029713 creator A5071418840 @default.
• W4207029713 date "2021-12-07" @default.
• W4207029713 modified "2023-09-24" @default.
• W4207029713 title "Group and Shuffle Convolutions for High-Resolution Semantic Segmentation" @default.
• W4207029713 cites W1970525044 @default.
• W4207029713 cites W2020028596 @default.
• W4207029713 cites W2044427921 @default.
• W4207029713 cites W2804047627 @default.
• W4207029713 cites W2966096374 @default.
• W4207029713 cites W2969548447 @default.
• W4207029713 cites W2975602169 @default.
• W4207029713 cites W2998168085 @default.
• W4207029713 cites W3003481805 @default.
• W4207029713 cites W3033959985 @default.
• W4207029713 cites W3035941285 @default.
• W4207029713 cites W3052130098 @default.
• W4207029713 cites W3092462775 @default.
• W4207029713 cites W3158776888 @default.
• W4207029713 cites W3172722332 @default.
• W4207029713 cites W3182640657 @default.
• W4207029713 cites W3189776055 @default.
• W4207029713 cites W3191527506 @default.
• W4207029713 doi "https://doi.org/10.1109/dasa53625.2021.9682288" @default.
• W4207029713 hasPublicationYear "2021" @default.
• W4207029713 type Work @default.
• W4207029713 citedByCount "0" @default.
• W4207029713 crossrefType "proceedings-article" @default.
• W4207029713 hasAuthorship W4207029713A5050837249 @default.
• W4207029713 hasAuthorship W4207029713A5054106321 @default.
• W4207029713 hasAuthorship W4207029713A5071418840 @default.
• W4207029713 hasConcept C108583219 @default.
• W4207029713 hasConcept C124504099 @default.
• W4207029713 hasConcept C138885662 @default.
• W4207029713 hasConcept C153180895 @default.
• W4207029713 hasConcept C154945302 @default.
• W4207029713 hasConcept C177264268 @default.
• W4207029713 hasConcept C199360897 @default.
• W4207029713 hasConcept C205649164 @default.
• W4207029713 hasConcept C2776401178 @default.
• W4207029713 hasConcept C2778755073 @default.
• W4207029713 hasConcept C31972630 @default.
• W4207029713 hasConcept C41008148 @default.
• W4207029713 hasConcept C41895202 @default.
• W4207029713 hasConcept C52622490 @default.
• W4207029713 hasConcept C58640448 @default.
• W4207029713 hasConcept C81363708 @default.
• W4207029713 hasConcept C89600930 @default.
• W4207029713 hasConceptScore W4207029713C108583219 @default.
• W4207029713 hasConceptScore W4207029713C124504099 @default.
• W4207029713 hasConceptScore W4207029713C138885662 @default.
• W4207029713 hasConceptScore W4207029713C153180895 @default.
• W4207029713 hasConceptScore W4207029713C154945302 @default.
• W4207029713 hasConceptScore W4207029713C177264268 @default.
• W4207029713 hasConceptScore W4207029713C199360897 @default.
• W4207029713 hasConceptScore W4207029713C205649164 @default.
• W4207029713 hasConceptScore W4207029713C2776401178 @default.
• W4207029713 hasConceptScore W4207029713C2778755073 @default.
• W4207029713 hasConceptScore W4207029713C31972630 @default.
• W4207029713 hasConceptScore W4207029713C41008148 @default.
• W4207029713 hasConceptScore W4207029713C41895202 @default.
• W4207029713 hasConceptScore W4207029713C52622490 @default.
• W4207029713 hasConceptScore W4207029713C58640448 @default.
• W4207029713 hasConceptScore W4207029713C81363708 @default.
• W4207029713 hasConceptScore W4207029713C89600930 @default.
• W4207029713 hasFunder F4320309480 @default.
• W4207029713 hasFunder F4320322699 @default.
• W4207029713 hasFunder F4320329006 @default.
• W4207029713 hasLocation W42070297131 @default.
• W4207029713 hasOpenAccess W4207029713 @default.
• W4207029713 hasPrimaryLocation W42070297131 @default.
• W4207029713 hasRelatedWork W11691867 @default.
• W4207029713 hasRelatedWork W13815759 @default.
• W4207029713 hasRelatedWork W14569654 @default.
• W4207029713 hasRelatedWork W1469282 @default.
• W4207029713 hasRelatedWork W3060714 @default.
• W4207029713 hasRelatedWork W3542719 @default.
• W4207029713 hasRelatedWork W4033802 @default.
• W4207029713 hasRelatedWork W4262719 @default.
• W4207029713 hasRelatedWork W6212567 @default.
• W4207029713 hasRelatedWork W6930659 @default.
• W4207029713 isParatext "false" @default.
• W4207029713 isRetracted "false" @default.
• W4207029713 workType "article" @default.