Unsupervised Sparse, Nonnegative, Low Rank Dictionary Learning for Detection of Driver Cell Phone Usage

Design of an Efficient Distracted Driver Detection System: Deep Learning Approaches

“…The models in this paper have been compared based on their train loss, test loss, and validation loss. The obtained results are compared with C.Huang et al [21], F.Omerustaoglu et al [22], and K.Roy et al [24] which are available in the literature.…”

“…The results obtained with the proposed DDDS method are compared with the results of C. Huang et al [21], F.Omerustaoglu et al [22], and K.Roy et al [24] in the next part of the sub-section. The results obtained using the DDDS methods are compared with the results of C. Huang et al [21] and summarized in Table 3.…”

“…The proposed Resnet-50 with random rotation and brightness train set augmentation, blur augmentation, and skin segmentation to achieve a F1-score of 0.662, which is 15% more than Resnet-50. K.Roy et al [24], to detect driver distraction because of mobile usage while driving, designed an unsupervised learning method called a low rank sparse non-negative dictionary (LRSNND). With the LRSNND method, an accuracy of 77.19% was obtained with the state farm distraction driving dataset.…”

“…The last comparison is performed with K.Roy et al [24] and summarized in Table 5. K.Roy et al [24] detected mobile usage while driving using an unsupervised learning method called LRSNND.…”

“…The last comparison is performed with K.Roy et al [24] and summarized in Table 5. K.Roy et al [24] detected mobile usage while driving using an unsupervised learning method called LRSNND. An accuracy of 77.19% is achieved with mobile usage classes of the state farm distraction driving dataset.…”