Sushish Baral
Sushish Baral has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- May Phu Paing
Department of Biomedical Engineering, School of Engineering, King Mongkut's Institute of Technology Ladkrabang, Bangkok, Thailand
3 shared publications - Nabin Sharma
Department of Robotics and AI, School of Engineering, King Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520, Thailand
2 shared publications - Rathachai Chawuthai
Department of Computer Engineering, School of Engineering, King Mongkuts Institute of Technology Ladkrabang, Bangkok, Thailand
2 shared publications Jagannath Aryal
1 shared publicationNabin Sharma
1 shared publication- Thinam Tamang
Madan Bhandari Memorial College, New Baneshwor, Kathmandu 44600, Nepal
1 shared publication
Latest Publications
- Open Access
- 12 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Remote Sensing Image Classification Using Transfer Learning Based Convolutional Neural Networks: An Experimental Overview
Image classification is one of the prominent tasks influencing the rapid advancement in computer vision using deep learning models such as convolutional neural networks, analogously gaining remarkable attention in remote sensing scene classification. However, the systematic comparability of literature concerning well-established large-scale or application-dependent datasets and models still requires improvement. This article intends to introduce the five high-capacity convolutional neural network architectures (VGG, ResNet, DenseNet, SqueezeNet, and AlexNet) and their variations with ten large-scale and small- scale datasets to provide a one-to-one experimentation summary. The observations were noted for comparison by keeping the configurations ideal in each trial. The performance of the models is depicted and discussed, with a focus on the number of flops, accuracy, F1-score, precision, recall, and AUCROC. This experiment validates the remote sensing image classification benchmark and provides insight for researchers and practitioners based on the summary and discussion for choosing the optimal model by comparing performance and the number of required floating point operations. Finally, the advanced optimization techniques of Normalization, Mixed-up augmentation, and Label smoothing are introduced to the best-chosen model in the case of each dataset.
- 12 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessRemote Sensing Image Classification Using Transfer Learning Based Convolutional Neural Networks: An Experimental Overview
Image classification is one of the prominent tasks influencing the rapid advancement in computer vision using deep learning models such as convolutional neural networks, analogously gaining remarkable attention in remote sensing scene classification. However, the systematic comparability of literature concerning well-established large-scale or application-dependent datasets and models still requires improvement. This article intends to introduce the five high-capacity convolutional neural network architectures (VGG, ResNet, DenseNet, SqueezeNet, and AlexNet) and their variations with ten large-scale and small- scale datasets to provide a one-to-one experimentation summary. The observations were noted for comparison by keeping the configurations ideal in each trial. The performance of the models is depicted and discussed, with a focus on the number of flops, accuracy, F1-score, precision, recall, and AUCROC. This experiment validates the remote sensing image classification benchmark and provides insight for researchers and practitioners based on the summary and discussion for choosing the optimal model by comparing performance and the number of required floating point operations. Finally, the advanced optimization techniques of Normalization, Mixed-up augmentation, and Label smoothing are introduced to the best-chosen model in the case of each dataset.
- Top Cited PapersOpen Access
Parking Time Violation Tracking Using YOLOv8 and Tracking Algorithms
The major problem in Thailand related to parking is time violation. Vehicles are not allowed to park for more than a specified amount of time. Implementation of closed-circuit television (CCTV) surveillance cameras along with human labor is the present remedy. However, this paper presents an approach that can introduce a low-cost time violation tracking system using CCTV, Deep Learning models, and object tracking algorithms. This approach is fairly new because of its appliance of the SOTA detection technique, object tracking approach, and time boundary implementations. YOLOv8, along with the DeepSORT/OC-SORT algorithm, is utilized for the detection and tracking that allows us to set a timer and track the time violation. Using the same apparatus along with Deep Learning models and algorithms has produced a better system with better performance. The performance of both tracking algorithms was well depicted in the results, obtaining MOTA scores of (1.0, 1.0, 0.96, 0.90) and (1, 0.76, 0.90, 0.83) in four different surveillance data for DeepSORT and OC-SORT, respectively.
Top Cited PapersOpen AccessParking Time Violation Tracking Using YOLOv8 and Tracking Algorithms
The major problem in Thailand related to parking is time violation. Vehicles are not allowed to park for more than a specified amount of time. Implementation of closed-circuit television (CCTV) surveillance cameras along with human labor is the present remedy. However, this paper presents an approach that can introduce a low-cost time violation tracking system using CCTV, Deep Learning models, and object tracking algorithms. This approach is fairly new because of its appliance of the SOTA detection technique, object tracking approach, and time boundary implementations. YOLOv8, along with the DeepSORT/OC-SORT algorithm, is utilized for the detection and tracking that allows us to set a timer and track the time violation. Using the same apparatus along with Deep Learning models and algorithms has produced a better system with better performance. The performance of both tracking algorithms was well depicted in the results, obtaining MOTA scores of (1.0, 1.0, 0.96, 0.90) and (1, 0.76, 0.90, 0.83) in four different surveillance data for DeepSORT and OC-SORT, respectively.
- Open Access
- 11 May 2023
- Published by MDPI AG in Preprints.org
Parking Time Violation Tracking using Yolov8 and DeepSORT
In Thailand, parking time violation is a major problem, especially for mini-marts. At present the task of detecting parking time violation is mainly conducted manually using Closed-Circuit Television (CCTV). This method requires additional human labour to track incoming and outgoing vehicles. Therefore, low cost time violation tracking is needed. To the best of our knowledge, there has not been any research for parking violation detection and tracking conducted for parking time limits. This paper introduces a novel parking time violation detection algorithm using the Yolov8 and DeepSORT tracking algorithms to track vehicles in consecutive frames. The presented parking violation tracking algorithm can provide a guideline for research in parking time violation detection.
- 11 May 2023
- Published by MDPI AG in Preprints.org
Open AccessParking Time Violation Tracking using Yolov8 and DeepSORT
In Thailand, parking time violation is a major problem, especially for mini-marts. At present the task of detecting parking time violation is mainly conducted manually using Closed-Circuit Television (CCTV). This method requires additional human labour to track incoming and outgoing vehicles. Therefore, low cost time violation tracking is needed. To the best of our knowledge, there has not been any research for parking violation detection and tracking conducted for parking time limits. This paper introduces a novel parking time violation detection algorithm using the Yolov8 and DeepSORT tracking algorithms to track vehicles in consecutive frames. The presented parking violation tracking algorithm can provide a guideline for research in parking time violation detection.
- Open Access
- 22 November 2022
- Published by MDPI AG in Diagnostics
- Vol. 12 (12) , 2903
Classification of White Blood Cells: A Comprehensive Study Using Transfer Learning Based on Convolutional Neural Networks
White blood cells (WBCs) in the human immune system defend against infection and protect the body from external hazardous objects. They are comprised of neutrophils, eosinophils, basophils, monocytes, and lymphocytes, whereby each accounts for a distinct percentage and performs specific functions. Traditionally, the clinical laboratory procedure for quantifying the specific types of white blood cells is an integral part of a complete blood count (CBC) test, which aids in monitoring the health of people. With the advancements in deep learning, blood film images can be classified in less time and with high accuracy using various algorithms. This paper exploits a number of state-of-the-art deep learning models and their variations based on CNN architecture. A comparative study on model performance based on accuracy, F1-score, recall, precision, number of parameters, and time was conducted, and DenseNet161 was found to demonstrate a superior performance among its counterparts. In addition, advanced optimization techniques such as normalization, mixed-up augmentation, and label smoothing were also employed on DenseNet to further refine its performance.
- 22 November 2022
- Published by MDPI AG in Diagnostics
- Vol. 12 (12) , 2903
Open AccessClassification of White Blood Cells: A Comprehensive Study Using Transfer Learning Based on Convolutional Neural Networks
White blood cells (WBCs) in the human immune system defend against infection and protect the body from external hazardous objects. They are comprised of neutrophils, eosinophils, basophils, monocytes, and lymphocytes, whereby each accounts for a distinct percentage and performs specific functions. Traditionally, the clinical laboratory procedure for quantifying the specific types of white blood cells is an integral part of a complete blood count (CBC) test, which aids in monitoring the health of people. With the advancements in deep learning, blood film images can be classified in less time and with high accuracy using various algorithms. This paper exploits a number of state-of-the-art deep learning models and their variations based on CNN architecture. A comparative study on model performance based on accuracy, F1-score, recall, precision, number of parameters, and time was conducted, and DenseNet161 was found to demonstrate a superior performance among its counterparts. In addition, advanced optimization techniques such as normalization, mixed-up augmentation, and label smoothing were also employed on DenseNet to further refine its performance.