Leveraging Deep Learning for Inattentive Driving Behavior with In-Vehicle Cameras
| Autor: | Arafat Al-Dweik, Radu Muresan, Shanhong Liu |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
050210 logistics & transportation
Focus (computing) business.industry Computer science Deep learning 010401 analytical chemistry 05 social sciences Feature extraction Body movement Cloud computing 01 natural sciences Convolutional neural network 0104 chemical sciences Phone Human–computer interaction 0502 economics and business Artificial intelligence business Inattentive Driving |
| Zdroj: | ISNCC |
| DOI: | 10.1109/isncc49221.2020.9297247 |
| Popis: | Driver inattentiveness during driving is a major cause in road accidents. In general, the inattentiveness is due to external distractions that change driver’s focus from driving to non-driving activities. Hence, it is of imperative importance to alert drivers of their inattentiveness behaviors to prevent any possible accident. This paper investigates the inattentiveness behaviors such as texting over the phone, talking on the phone, tuning the radio player, eating and drinking, turn behind, makeup, and talking to passengers. We consider a car system that has a camera installed such that the camera will be capable of capturing the driver’s body movement. Convolutional neural network (CNN) is used to extract image features from the camera video stream and perform the classification. We present performance results of model development, model loaded into vehicle system, and model updated on custom cloud dataset. The cross-validation evaluation indicates that our proposed approach offers a simple, reliable, low-cost and high in-vehicle model accuracy (> 92%) solution in detecting the driver’s inattentiveness problem during driving. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|