A Novel Faster RCNN with ODN-Based Rain Removal Technique

During rainy times, the impact of outdoor vision systems gets considerably decreased owing to visibility barrier, distortion, and blurring instigated by raindrops. So, it is essential eradicate from images for ensuring reliability system. To achieve this, several rain removal studies have been performed in recent days. In this view, paper presents a new Faster Region Convolutional Neural Network (Faster RCNN) with Optimal Densely Connected Networks (DenseNet)-based technique called FRCNN-ODN. The presented involves weighted mean filtering (WMF) applied as denoising technique, which helps boost quality input image. addition, RCNN used detection that comprises region proposal network (RPN) Fast model. RPN generates high proposals are exploited detect drops. Also, DenseNet model utilized baseline generate feature map. Moreover, sparrow search optimization algorithm (SSOA) choose hyperparameters namely learning rate, batch size, momentum, weight decay. An extensive experimental validation process highlight effectual outcome FRCNN-ODN investigated results respect dimensions. method produced higher UIQI 0.981 image 1. Furthermore, on 2, achieved maximum 0.982. 0.998 3. simulation showcased superior (Optimal Networks) existing methods terms distinct measures.