Tarik Reza Toha

Ph.D. Student

UNC Computer Science

Chapel Hill, NC 27514

Email: ttoha12@cs.unc.edu

Research Interests
  1. Sensing Systems
  2. Computer Vision
  3. Deep Learning

Hello! I'm Tarik Reza Toha, a third-year Ph.D. student in Computer Science at the University of North Carolina at Chapel Hill, advised by Prof. Shahriar Nirjon. I work at the intersection of deep learning and sensing systems, integrating signal processing with generative models across mmWave, WiFi, and wearable modalities, for privacy-preserving applications.

Before starting my Ph.D., I worked as a Lecturer in the Department of Textile Machinery Design and Maintenance at the Bangladesh University of Textiles. I received both my Bachelor's and Master's degrees in Computer Science and Engineering from the Bangladesh University of Engineering and Technology, where I was advised by Prof. A. B. M. Alim Al Islam.

Educational Background

University of North Carolina at Chapel Hill
  • Ph.D. in Computer Science, August 2023 - Present
    • - Advisor: Prof. Shahriar Nirjon
Bangladesh University of Engineering and Technology
  • M.Sc. in Computer Science and Engineering, October 2017 - September 2021
    • - Advisor: Prof. A. B. M. Alim Al Islam
Bangladesh University of Engineering and Technology
  • B.Sc. in Computer Science and Engineering, February 2013 - September 2017
    • - Advisor: Prof. A. B. M. Alim Al Islam

Research Experience

University of North Carolina at Chapel Hill
  • Graduate Research Assistant, August 2023 - Present
    • - Inertial Video Generation: Running
    • - mmWave Anomaly Detection: SenSys 2026
    • - mmWave-based People Counting: EWSN 2025
Bangladesh University of Engineering and Technology
  • Graduate Research Assistant, October 2017 - September 2021
    • - Embedded Object Detection: ICDM 2021
    • - Dense Crowd Localization: ASOC 2022
    • - Green MapReduce Clusters: ICC 2018, TPDS 2021

Teaching Experience

University of North Carolina at Chapel Hill
  • Graduate Teaching Assistant, August 2023 - Present
    • - Mobile Computing Systems: Android, iOS
    • - Files and Databases: SQLite, Python
Bangladesh University of Textiles
  • Lecturer, December 2018 - Present
    • - Programming Languages: C, MATLAB, Python
    • - Database and Information Systems: MySQL, Laravel
    • - Embedded Systems: Arduino, Proteus
    • - Data Analytics: Python, Machine Learning

Selected Projects

Sensing Systems
  1. Inertial Video Generation: Developed Imu2Vid, a video generation pipeline that generates realistic human motion videos from wearable sensors using a pose-guided multimodal diffusion model. Integrated human pose estimation from inertial motion signals, avatar animation using a Diffusion Transformer, and physics-based video refinement.
    Skills: Wearable Sensing, Diffusion Transformers, HPC, PyTorch
  2. mmWave Anomaly Detection: Developed mmAnomaly, a multimodal framework combining mmWave radar and RGBD input for privacy-aware anomaly detection in cluttered indoor settings. Integrated a ResNet-based visual feature extractor, a text-guided stable diffusion model for signal synthesis, and a ViT-based comparison module for anomaly localization.
    Skills: Multimodal Sensing, Generative Models, Vision Transformers, HPC, PyTorch
  3. mmWave-based People Counting: Developed mmCounter, a mmWave sensing system for counting stationary individuals by capturing low-frequency physiological signals from breathing and micro-movements. Used FFT and CFAR for motion isolation, iterative ICA for spatial signal separation, and a Vision Transformer for signal-to-person mapping.
    Skills: mmWave Sensing, Signal Processing, Vision Transformers, PyTorch
  4. WiFi-based Activity Recognition: Developed WiHAR, a WiFi sensing system for human activity recognition in privacy-sensitive environments using WiFi Channel State Information (CSI). Processed CSI amplitude data to generate Channel Frequency Responses and classified activities using a Vision Transformer model.
    Skills: WiFi Sensing, Signal Processing, Vision Transformers, PyTorch
Computer Vision
  1. Embedded Object Detection: Designed DhakaNet, a lightweight deep learning model for real-time object detection on edge devices in traffic environments. Extended YOLOv5 by enhancing the backbone and neck with efficient architectural blocks and integrating a custom multi-scale attention module to boost feature extraction under resource constraints.
    Skills: Edge AI, Object Detection, PyTorch, Raspberry Pi, Laravel, MySQL
  2. Dense Crowd Localization: Designed LC-Net, a deep convolutional network for accurate person localization in ultra-dense crowd scenarios such as Hajj and urban gatherings. Incorporated residual connections to mitigate vanishing gradients and dilated convolutions to retain spatial context under occlusion and distortion.
    Skills: Deep Learning, Convolutional Neural Networks, Data Augmentation, Keras, AWS, Raspberry Pi, Laravel, MySQL
Distributed Systems
  1. Green MapReduce Clusters: Developed GMC, a machine learning–based framework that reduces total energy consumption in MapReduce clusters by jointly optimizing computational and cooling energy. Predicted job-level energy usage and dynamically adjusted cluster size using models trained on real-world workload data.
    Skills: Distributed Systems, Machine Learning, Hadoop, SimGrid, Arduino, Java, MySQL
Miscellaneous
  1. Digital Pre-Distorter: Developed an LSTM–based digital predistortion system via modeling a Power Amplifier (PA) and training a digital predistorter with the frozen PA model to enhance RF power amplifier linearity.
    Skills: PA Modeling, DPD Learning, LSTM, PyTorch
  2. Heart Rate Monitor: Built a standalone heart rate monitoring system using an Atmega328P microcontroller, a heart rate sensor, and an OLED display to detect beats, estimate BPM, and visualize IR signals in real time on a custom PCB board.
    Skills: Arduino, Heart Rate Sensor, OLED Display, PCB Design
  3. Bluetooth Messaging App: Developed a macOS Peripheral app that advertises a custom GATT service with RX/TX characteristics, and an iOS Central app that discovers, connects, and manages subscriptions for reliable two-way messaging over Bluetooth Low Energy (BLE) communication.
    Skills: macOS Development, iOS Development, Bluetooth Low Energy
  4. Secured Messaging App: Integrated an AES-GCM encryption with native C++ padding via JNI into an XMPP-based Android messaging app to enhance end-to-end security in message handling and transmission.
    Skills: Android Development, Cryptography, C++

Selected Publications

Conference Papers
  1. T. R. Toha, S.-J. L. Lu, M. Monjur, and S. Nirjon, "mmAnomaly: Leveraging Visual Context for Robust Anomaly Detection in the Non-Visual World with mmWave Radar," in Proceedings of the 24th ACM/IEEE International Conference on Embedded Artificial Intelligence and Sensing Systems (SenSys). Saint-Malo, France: ACM, May 2026, accepted. [arXiv]. Available: https://doi.org/10.48550/arXiv.2604.00382
  2. T. R. Toha, S.-J. L. Lu, and S. Nirjon, "mmCounter: Static People Counting in Dense Indoor Scenarios Using mmWave Radar," in Proceedings of the 22nd International Conference on Embedded Wireless Systems and Networks (EWSN). Leuven, Belgium: ACM, September 2025, accepted. [arXiv]. Available: https://doi.org/10.48550/arXiv.2512.10357
  3. T. R. Toha, M. Rahaman, S. I. Salim, M. Hossain, A. M. Sadri, and A. B. M. A. A. Islam, "DhakaNet: Unstructured Vehicle Detection using Limited Computational Resources," in Proceedings of the 21st IEEE International Conference on Data Mining (ICDM). Auckland, New Zealand: IEEE, December 2021, pp. 1367–1372. [Online]. Available: https://doi.org/10.1109/ICDM51629.2021.00172
  4. T. R. Toha, M. M. R. Lunar, A. S. M. Rizvi, N. Nurain, and A. B. M. A. A. Islam, "GMC: Greening MapReduce Clusters Considering both Computational Energy and Cooling Energy," in Proceedings of the 52nd IEEE International Conference on Communications (ICC). Kansas City, MO, USA: IEEE, May 2018, pp. 1–6. [Online]. Available: https://doi.org/10.1109/ICC.2018.8422113
  5. T. R. Toha, A. M. Ishmam, M. H. Islam, M. A. A. Maruf, S. S. Nandi, A. Chakraborty, S. Estyak, M. A. A. Alamin, and A. B. M. A. A. Islam, "An Approach Towards Greening the Digital Display System," in Proceedings of the 4th International Conference on Networking, Systems and Security (NSysS). Dhaka, Bangladesh: IEEE, December 2017, pp. 1–6. [Online]. Available: https://doi.org/10.1109/NSYSS2.2017.8267794
  6. T. R. Toha, S. Estyak, T. A. Khan, T. Chakraborty, and A. B. M. A. A. Islam, "Sparse Mat: A Tale of Devising A Low-Cost Directional System for Pedestrian Counting," in Proceedings of the 3rd International Conference on Networking, Systems and Security (NSysS). Dhaka, Bangladesh: IEEE, January 2017, pp. 21–29. [Online]. Available: https://doi.org/10.1109/NSysS.2017.7885796 [Best Paper Award]
Journal Papers
  1. M. M. Mushfiq, T. R. Toha, S. I. Salim, A. Mostak, M. Rahaman, N. A. Al-Nabhan, A. M. Sadri, and A. B. M. A. A. Islam, "To Lane or Not to Lane? – Comparing On-Road Experiences in Developing and Developed Countries Using a New Simulator RoadBird," IEEE Transactions on Intelligent Transportation Systems (TITS), vol. 25, no. 8, pp. 8486–8498, June 2024. [Online]. Available: https://doi.org/10.1109/TITS.2024.3406731
  2. T. R. Toha, N. A. Al-Nabhan, S. I. Salim, M. Rahaman, U. Kamal, and A. B. M. A. A. Islam, "LC-Net: Localized Counting Network for Extremely Dense Crowds," Applied Soft Computing Journal (ASOC), Elsevier, vol. 123, p. 108930, July 2022. [Online]. Available: https://doi.org/10.1016/j.asoc.2022.108930
  3. T. R. Toha, A. S. M. Rizvi, J. Noor, M. A. Adnan, and A. B. M. A. A. Islam, "Towards Greening MapReduce Clusters Considering both Computation Energy and Cooling Energy," IEEE Transactions on Parallel and Distributed Systems (TPDS), vol. 32, no. 4, pp. 931–942, April 2021. [Online]. Available: https://doi.org/10.1109/TPDS.2020.3029724