Fahad Majeed 
PhD Scholar | Computer Vision Researcher | Sports Video Analyst
I am a PhD Scholar in Computer Vision with over 8 years of cross-disciplinary experience in academia and industry. My core expertise lies in designing deep learning-based pipelines for solving complex challenges in sports video analytics, and large-scale data processing. I develop end-to-end systems that combine computer vision, graph theory, AI, and big data technologies to deliver robust, scalable, and high-impact solutions. My research pushes the boundaries of AI for human activity understanding, real-time tracking, pose estimation, event detection, and instance segmentation, particularly in domains like sports performance analysis.
🎓 Areas of Interest
- 📌 Thesis Focus: Multi-modal Deep Learning & Graph-based Reasoning for Player Behavior Understanding in Sports Videos
- 🧠 Specializations: Object Detection, Instance Segmentation, Graph Neural Networks (GNN), Feature Pyramid Networks (FPN), Temporal Modeling
- ⚽ Sports AI: Real-time player tracking, tactical analytics, multi-camera calibration, ball trajectory prediction
- ☁️ Big Data: Distributed video annotation pipelines using Apache Spark, Pig Latin, and cloud-based GPU processing
📈 Key Technical Skills
🧠 Machine Learning / Deep Learning
- PyTorch, TensorFlow, Scikit-learn, Keras, Detectron2, MMDetection
- CNNs, RNNs, Transformers, Attention Mechanisms, SepFormer
- Graph Neural Networks (GCN, GAT, ST-GCN)
📊 Computer Vision
- Object Detection (YOLOv3/5/7/8/9/10/11/12/YOLOE, Faster R-CNN, Mamba-YOLO)
- Instance Segmentation (Mask R-CNN)
- Multi-object Tracking (SORT, OC-SORT, Deep SORT, ByteTrack, and BoTSORT)
- Pose Estimation (OpenPose, HRNet, Mediapipe)
🔗 Graph Learning & Temporal Analysis
- GCNConv, EdgeConv, ST-GCN
- Spatio-temporal adjacency modeling
- Action prediction from pose graph sequences
⚙️ Tools & Platforms
- Python, MATLAB, Bash, Git, Docker, LaTeX, REST APIs
- Jupyter Notebooks, PyCharm, VSCode
- Streamlit, Flask
- Kaggle, Google Colab Pro, NVIDIA CUDA
📚 Projects
⚽ Intelligent Soccer Analytics Framework
A modular pipeline integrating:
- Deep instance segmentation to detect players
- Graph-based spatial-temporal reasoning to model interactions
- Pose-based action recognition and team formation detection
👉 Used in performance evaluation and tactical insights.
📹 Real-time Multi-camera Video Analytics
- Camera calibration and homography estimation
- Player re-identification using feature embedding + appearance descriptors
- Action tube generation and spatio-temporal event localization
👉 Built for large-scale soccer datasets with over 1000+ video hours.
🧪 Publications
- “ReST: High-Precision Soccer Player Tracking via Motion Vector Segmentation “, 20th VISAPP Conference, 2025, Porto, Portugal.
- “MV-Soccer: Motion-Vector Augmented Instance Segmentation for Soccer Player Tracking”, 10th CVSports Workshops (CVPRW), 2024, Seattle, USA.
🧬 Datasets
- SoccerNet, SoccerNet-Tracking, FIFA World Cup Data, OpenPose, BraTS
- Annotated custom datasets using CVAT, LabelMe, Makesense.ai, and Roboflow.
💼 Professional Experience
-
Teaching & Mentorship:
Taught undergraduate courses in Computer Science (BS level)
Supervised undergraduate research projects and thesis in CV/AI/IoT/ML domains -
Industry Projects:
Worked on AI-driven medical diagnostics, surveillance systems, and sports tech for startups and research labs.
—
📫 Contact
📧 fahad_majeed@yahoo.com
📍 Education City, Qatar Foundation, Doha, Qatar
🌐
“Research is the art of seeing what everyone else has seen, but thinking what no one else has thought.”
– Fahad Majeed