Enhanced Deep Smart Intelligence Based Driver Drowsiness Detection Using Optimal Scalar Vector Feature Selection with Convolution Neural Network

Driver drowsiness detection is important aspect for avoiding accidental risk to protect safeguard for al type of transport. Through artificial technology take effect for identifying automated driver behavior detection using image screening and mentor the driver face recognitions to identifying the drowsiness impacts. Traditional models uses various aspects to find the structural points to identify the risk of evaluation, but increasing feature dimension. The pitfall of eye lids and drowsy behaviors are so tedious to recognize the changes due to segmentation failures which leads poor detection accuracy, precision ad recall rate provide higher false negatives. To resolve this problem. We propose a enhanced deep smart intelligence based drowsiness detection using optimal scalar vector feature selection (OSVFS) with Resnet 50-convolution neural network (RSN-CNN). Initially the median filer is applied to normalize the image scaling and marginalize the feature ideal margins of face lids presence to improve the contrasting. Then Eye Pit point aspect segmentation is applied to identify the face cover eyelid cover region to mark the closeness of lid space count. Further the Deep vectorized scalar feature selection is applied to extract the drowsiness lid closeness of behavior impact rate. The behavioral aspects of drowsiness of feature margins are validated with absolute scalar margin of closing eye lid counts to identify the drowsiness pattern. Finally the deep convolution neural network is applied to categories feature margins according to feature weight with the class by range of drowsiness presence. Along with the class weights, drivers are alert to avoid the risk to make safe driving system. The proposed system produce high performance by identifying behavioral properties by correcting the feature segregation to increase the false [positive as well in higher precision rate, really rate and higher detection accuracy compared to the other system.