| Autor: |
Dasgupta, Madhuchhanda, Bandyopadhyay, Oishila, Chatterji, Sanjay |
| Předmět: |
|
| Zdroj: |
Multimedia Tools & Applications; Feb2023, Vol. 82 Issue 4, p5857-5877, 21p |
| Abstrakt: |
In a country like India with high density of population, motorcycle is one of the common and viable mode of transport. It is observed that many motorcyclists refrain from wearing helmets while driving. This results in fatal road accidents every year. In crowded roads and highways, it becomes difficult for the police to identify such cases and to take necessary actions. These traffic rule violators can be detected by analysing the traffic videos of surveillance camera. The main objective of this work is to detect the helmetless motorcyclists (and pillion riders) and generate appropriate video caption to help the traffic authority to take fast action against the rule violators. The system can also detect helmetless multiple riders and child rider cases from the video captions. A deep neural network based approach is proposed to generate the video captions for motorcycle riders from surveillance video analysis. In the proposed encoder-decoder based model, Convolutional Neural Network (CNN) along with optical flow guided approach are used for visual feature extraction in encoder part. In the decoder part, Recurrent Neural Network (RNN) based Long-Short-Term-Memory (LSTM) with Soft Attention (SA) technique is applied to achieve best result for video caption generation. The effectiveness of the proposed approach is evaluated by computing BiLingual Evaluation Understudy (BLEU) and Metric for Evaluation of Translation with Explicit Ordering (METEOR) metrices. The extensive experimental results show that the proposed method outperforms other state-of-the-art methods. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink