Publications

2024:

Nizamuddin Maitlo; Nooruddin Noonari; Sajid Ahmed Ghanghro; Sathishkumar Duraisamy; Fayaz Ahmed
arXiv [paper]
Our research introduces a novel approach to enhance the robustness of color detection in varying lighting conditions by leveraging a Convolutional Neural Network (CNN). Specifically, we maintain the integrity of image segmentation using edge detection techniques to isolate objects and then incorporate a CNN trained to detect object colors under diverse lighting conditions. This approach not only preserves the accuracy of object segmentation but also optimizes the color detection capabilities within the network, leading to significantly improved performance compared to existing methods.

Nizamuddin Maitlo; Nooruddin Noonari; Kaleem Arshid; Naveed Ahmed; Sathishkumar Duraisamy
arXiv [paper]
Our research introduces a novel approach to enhance the effectiveness and efficiency of indoor navigation for autonomous vehicles by leveraging a low-cost solution. Specifically, we retain the fundamental navigation capabilities of the base system and incorporate a monocular camera to address the challenges posed by GPS inaccuracy in indoor scenarios. Our solution, termed Affordable Indoor Navigation Solution (AINS), is designed to function without the need for extensive or power-inefficient sensors such as range finders, thereby maintaining cost-effectiveness. Through this approach, our method demonstrates superior performance over existing solutions, significantly reducing both estimation errors and time consumption.

Sajid Ahmed Ghanghro, Dr. Muhammad Ajmal Sawand Wajid Ahmed Channa, Ubaidulla alias Kashif, Muhammad Hanif Tunio, Kishor Kumar, Nooruddin Noonari

Nooruddin Noonari; Rahool Dembani; Nizamuddin Maitlo;
arXiv [paper]
Our research introduces a novel approach to enhance the effectiveness and efficiency of indoor navigation for autonomous vehicles by leveraging a low-cost solution. Specifically, we retain the fundamental navigation capabilities of the base system and incorporate a monocular camera to address the challenges posed by GPS inaccuracy in indoor scenarios. Our solution, termed Affordable Indoor Navigation Solution (AINS), is designed to function without the need for extensive or power-inefficient sensors such as range finders, thereby maintaining cost-effectiveness. Through this approach, our method demonstrates superior performance over existing solutions, significantly reducing both estimation errors and time consumption.

R Dembani, W Zheng Nooruddin Noonari
arXiv [paper]
Our study proposes a novel approach to enhance the accuracy and efficiency of facial expression clustering by employing a genetic algorithm. Specifically, we convert images into a binary format to enable the effective selection of related clusters through various phases of the genetic algorithm. The proposed method integrates a modified teacher learning-based optimization algorithm, which updates the population in each phase to identify the most representative features. We utilize a real dataset of facial expressions to validate our approach and compare its performance against existing models across different evaluation parameters. Our findings demonstrate that the proposed method significantly improves precision, recall, and accuracy in facial expression identification without requiring any training.