Working on a research collaboration project between ITU and the Electronics and Telecommunication Research Institute (ETRI) South Korea on 3D reconstruction. The group is currently focusing on color point cloud up-sampling and neural radiance field for 3D mesh reconstruction. 1. In the color point cloud up-sampling project, I actively contributed to the development of an attention-based feature extractor for patch-based point cloud up-sampling. By leveraging this approach, we aimed to enhance the quality and resolution of the up-sampled point clouds. I played a key role in designing and implementing the attention mechanism, resulting in improved accuracy and fidelity of the up-sampling process. 2. My involvement in the Neural Radiance Field (NeRF) project focused on extracting a smooth 3D mesh from volumetric representation using implicit surface information. Within this project, I actively collaborated with a team to develop innovative techniques that enabled the accurate reconstruction of complex surfaces. By leveraging the implicit surface information, we successfully obtained high-quality 3D meshes with improved smoothness and fidelity.

I actively contributed to the development of a human 3D pose estimation model and successfully mapped the motion onto the Unity engine for augmented reality applications. I utilized state-of-the-art pose estimation algorithms, such as VideoPose3D, and collaborated with a team to optimize the model's accuracy and robustness. As part of a collaborative team, I played a key role in training and deploying a 3D human mesh reconstruction model for a web application. We leveraged advanced deep learning techniques, to achieve accurate and detailed reconstructions of human meshes from 2D input data. I actively contributed to the model training process, data preprocessing, and optimizing the network's performance. Our successfully deployed web app allowed users to interactively view and manipulate 3D human models.

• I trained and tested various state-of-the-art deep learning models for 3D object detection, and implemented pipelines in Pytorch, Numpy, and PCL. I utilized large-scale datasets such as the KITTI Dataset and employed evaluation metrics like 3D IoU to assess the models' performance.• I contributed to the development and deployment of a proprietary Road Surface Analysis Tool. This tool was designed to analyze road conditions and specifically detect damaged road patches.

I had the opportunity to work on diverse tasks related to object tracking, detection, and object counting using the Intel RealSense D435 camera. Within this role, I applied various computer vision techniques and algorithms to track objects in real time, detect their presence, and accurately count them.

A community startup that promotes Data science in Pakistan.