Publications & Research

3D-FFS: Faster 3D object detection with Focused Frustum Search in sensor fusion based networks

Resources: [Preprint], [Article]
Supervisor: Dr. Md. Shamsuzzoha Bayzid
Publication: Accepted at the 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2021).
Keywords: Deep Learning, 3D Computer Vision, Autonomous Vehicles
Description: In this work, we propose a novel approach to make sensor fusion based 3D object detection networks significantly faster using a class of computationally inexpensive heuristics. We worked on existing 3D object detection architectures and made adjustments that improve inference time, training time and performance.

Online Detection of Attentiveness of Students with Special Needs

Resources: [Article]
Supervisor: Dr. Mohammed Eunus Ali
Publication: Accepted at the 10th International Conference on Affective Computing & Intelligent Interaction (ACII 2022)
Keywords: : Assistive Technology, Computer Vision, Attentiveness Detection
Description: In this paper, we propose the first deep learning based attentiveness level prediction technique for students with ASD. Our model detects the behavior (e.g., unusual movement, gaze etc.) and activities from real-time videos and uses them as features to classify the attentiveness level as low, mid and high. Existing state-of-the-art techniques to detect the attentiveness level of typically developed students using gaze or facial expression cannot be trivially extended for students with ASD as they do not exhibit regular and consistent behavior. We collect video data belonging to different classes covering various types of activities over a long period, train our classifier, and run extensive experiments to validate the prediction performance. Our solution outperforms existing baselines by a large margin.

Large Scale Extractive News Summarization Benchmark Dataset in Bangla

Technology: Text Summarization, Natural Language Processing, Low Resource Languages, Benchmark Dataset
Supervisor: Dr. Mohammed Eunus Ali
Duration: March, 2021 - April, 2022
Description: Project to create the first large-scale high quality news summarization benchmark dataset in Bangla.
Responsibilities:

• Implemented multiple state of the art BERT and RNN based news summarization models for Bangla.
• Did extensive work in data preparation, data processing and data quality evaluation.

Exploring Oppurtunities for Online Education of Special Needs Individuals in Bangladesh

Technology: Human Computer Interaction, Accessibility, Qualitative Research
Duration: November, 2020 - April, 2022
Supervisor: Dr. Mohammed Eunus Ali
Description: Research project to identify problems faced by special needs children in online education.
Responsibilities:

• Did qualitative research by conducting interviews, surveys and need-finding.
• Developed deep learning models for detecting attentiveness, mood, actions, etc based on created dataset.

Faster 3D object detection with Focused Frustum Search in sensor fusion based networks

Technology: Deep Learning, 3D Computer Vision, Autonomous Vehicles
Duration: February, 2020 - March, 2021
Supervisor: Dr. Md. Shamsuzzoha Bayzid
Description: Research project exploring sensor fusion to combine RGB and 3D LIDAR data for object detection and localization. Responsibilities:

• Developed and implemented novel methodologies for improving runtime and accuracy of 3D object detection models.