A Computer Vision Based Approach forDriver Distraction Recognition Using Deep Learning and Genetic Algorithm Based Ensemble

As the proportion of road accidents increases each year, driver distraction continues to be an important risk component in traffic injuries and deaths. The distractions caused by increasing use mobile phones other wireless devices pose a potential safety. Our current study aims aid already existing techniques posture recognition improving performance classification problem. We present approach using genetic algorithm-based ensemble six independent deep neural architectures, namely, AlexNet, VGG-16, EfficientNet B0, Vanilla CNN, Modified DenseNet, InceptionV3 + BiLSTM. test it on two comprehensive datasets, AUC Distracted Driver Dataset, which our technique achieves accuracy 96.37%, surpassing previously obtained 95.98%, State Farm Distraction we attain 99.75%. 6-Model Ensemble gave inference time 0.024 seconds as measured machine with Ubuntu 20.04(64-bit) GPU GeForce GTX 1080.