Saravjeet Singh
Research Subjects
Saravjeet Singh has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Rishu Chhabra
Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura, India
2 shared publications - Jatin Arora
Chitkara University
1 shared publication - Neeru Sharma1 shared publication
- Ravneet Kaur
CGC University
1 shared publication - Rupali Gill
Chitkara University Institute of Engineering and Technology, Chitkara University, Chandigarh, Punjab, INDIA
1 shared publication
Latest Publications
- Open Access
- 18 November 2025
- Published by Springer Nature in Journal of Transformative Technologies and Sustainable Development
- Vol. 9 (1)
A Hybrid Extractive and Encoder-Decoder-Based Approach for Mitigating Hallucination in Automatic Text Summarization
- 18 November 2025
- Published by Springer Nature in Journal of Transformative Technologies and Sustainable Development
- Vol. 9 (1)
Open AccessA Hybrid Extractive and Encoder-Decoder-Based Approach for Mitigating Hallucination in Automatic Text Summarization
- 13 September 2023
- Published by Taylor & Francis
- p. 163-170
Deep-Learning-Based Road Surface Classification for Intelligent Vehicles
Intelligent vehicles (IVs) are key component of intelligent transportation systems (ITSs) and aim at improving the efficiency and convenience in transportation system by employing advanced technologies. With the rise in the count of road crashes, road surface information is a key factor in intercepting road accidents. In this chapter, we employ different deep learning techniques for classifying the road surfaces into asphalt, gravel, and concrete road. Different deep learning algorithms, CNN, LSTM, and InceptionV3, have been implemented and analysed using Mendeley road surface image dataset. It is evident from the results that the InceptionV3 algorithm yields highest training accuracy of 93% and validation accuracy of 64%. This is supported by the acceptable validation AUC of approximately 86%. The results from the road surface classification algorithm can be used by intelligent and autonomous vehicles to control the vehicle dynamics like speed and braking behaviour to improve the safety on roads.
- 13 September 2023
- Published by Taylor & Francis
- p. 163-170
Deep-Learning-Based Road Surface Classification for Intelligent Vehicles
Intelligent vehicles (IVs) are key component of intelligent transportation systems (ITSs) and aim at improving the efficiency and convenience in transportation system by employing advanced technologies. With the rise in the count of road crashes, road surface information is a key factor in intercepting road accidents. In this chapter, we employ different deep learning techniques for classifying the road surfaces into asphalt, gravel, and concrete road. Different deep learning algorithms, CNN, LSTM, and InceptionV3, have been implemented and analysed using Mendeley road surface image dataset. It is evident from the results that the InceptionV3 algorithm yields highest training accuracy of 93% and validation accuracy of 64%. This is supported by the acceptable validation AUC of approximately 86%. The results from the road surface classification algorithm can be used by intelligent and autonomous vehicles to control the vehicle dynamics like speed and braking behaviour to improve the safety on roads.
- 13 September 2023
- Published by Taylor & Francis
- p. 372-380
State of the Art of Quantum Teleportation and Quantum Key Distribution Methods
Due to high processing power, Quantum computing is having the potential to transform the computation process involved in different industries like transportation, security, drug development, and so on. Earlier quantum computing was the theoretical concept, but now with the advancement of hardware and software, many industries are providing cloud-based solutions for quantum processing. Based on the recent trends of quantum computing, this chapter provides brief survey of quantum computing techniques used in cybersecurity and information security. This chapter elucidates impact of quantum methods on security applications. This chapter highlights the recent state of art of quantum teleportation and quantum key distribution. The benefits, drawbacks, and methods used by various techniques are presented in this chapter. Furthermore, new research directions for quantum computing in cryptography are presented.
- 13 September 2023
- Published by Taylor & Francis
- p. 372-380
State of the Art of Quantum Teleportation and Quantum Key Distribution Methods
Due to high processing power, Quantum computing is having the potential to transform the computation process involved in different industries like transportation, security, drug development, and so on. Earlier quantum computing was the theoretical concept, but now with the advancement of hardware and software, many industries are providing cloud-based solutions for quantum processing. Based on the recent trends of quantum computing, this chapter provides brief survey of quantum computing techniques used in cybersecurity and information security. This chapter elucidates impact of quantum methods on security applications. This chapter highlights the recent state of art of quantum teleportation and quantum key distribution. The benefits, drawbacks, and methods used by various techniques are presented in this chapter. Furthermore, new research directions for quantum computing in cryptography are presented.
- 13 September 2023
- Published by Taylor & Francis
- p. 171-180
An Empirical Review of Potholes Classification Using Road Images
Road conditions have a huge impact on transportation and safe journey. For the better and easy prediction of potholes/road conditions many automatic detection techniques exist. In this chapter, transfer-learning-based techniques are used to classify road conditions. Road images numbering 1,719 were collected for this experiment. Visual Geometry Group (VGG) 16 and 19 models were implemented for the pothole classifications. The number of parameters derived using the VGG16 and VGG19 were 153,331,715 and 158,641,411, respectively. Training and testing data were divided according to 80:20 ratio. Models were trained on 30 epochs and according to performed experiment, VGG19 provided maximum classification accuracy of 93% with a 5% validation loss. VGG16 provided maximum accuracy of 92% with 2.11% of validation loss.
- 13 September 2023
- Published by Taylor & Francis
- p. 171-180
An Empirical Review of Potholes Classification Using Road Images
Road conditions have a huge impact on transportation and safe journey. For the better and easy prediction of potholes/road conditions many automatic detection techniques exist. In this chapter, transfer-learning-based techniques are used to classify road conditions. Road images numbering 1,719 were collected for this experiment. Visual Geometry Group (VGG) 16 and 19 models were implemented for the pothole classifications. The number of parameters derived using the VGG16 and VGG19 were 153,331,715 and 158,641,411, respectively. Training and testing data were divided according to 80:20 ratio. Models were trained on 30 epochs and according to performed experiment, VGG19 provided maximum classification accuracy of 93% with a 5% validation loss. VGG16 provided maximum accuracy of 92% with 2.11% of validation loss.