Hypertuned Convolutional Neural Network Residual Model Based Content Based Image Retrival System
Aman Singh, Amit Rai Dixit, Brajesh Kumar Singh
Abstract
Content-based Image Retrieval (CBIR) Framework is to find related photographs in an enormous data set. The ordinary technique is to remove a few significant qualities from the question image and recover them. Images with a comparative arrangement of qualities are recovered with high similitude scores. Framework’s prosperity and undeniable level attributes are important to close the semantic hole. In this paper two CNN models, ResNet50 and VGG16 have been considered for an enormous image order issue. Hyperparameter tuning and execution assessment is performed on the CINIC-10 dataset.
Topics & Concepts
Convolutional neural networkComputer scienceHyperparameterResidualImage (mathematics)Artificial intelligenceImage retrievalContent-based image retrievalSet (abstract data type)Pattern recognition (psychology)SimilitudeContent (measure theory)MathematicsAlgorithmMathematical analysisProgramming languageImage Retrieval and Classification TechniquesAdvanced Image and Video Retrieval TechniquesAdvanced Image Fusion Techniques