This page highlights selected research projects, industrial systems, and open-source contributions spanning Explainable AI, Computer Vision, and Anomaly Detection.

🤖 ADVIS-DistriMuSe-SR — Safe Interaction with Robots

Project Page GitHub Repo

A real-time Anomaly Detection and Visual Intelligence System (ADVIS) developed for the DistriMuSe Use Case 3: Safe Interaction with Robots. The system monitors industrial safety areas in collaborative robotics environments using VAE/VAE-GAN models, ROS2 image streams, threshold calibration, and RuleX-compatible alert publishing.

🔬 Main Contributions

• Safety-area-based anomaly detection for RoboArm, ConvBelt, PLeft, and PRight
• VAE/VAE-GAN training and reconstruction-based anomaly scoring
• Threshold calibration for robust detection
• ROS2 live inference from camera topics
• GUI-based inspection with anomaly maps, reconstructions, and timeline visualization
• RuleX-compatible message publishing for industrial integration

📊 Achievements

• ~12.5 FPS real-time inference
• 99.61% accuracy
• 95.1% recall
• 90.9% F1-score

• Tested in Smart Robotics / DistriMuSe integration scenarios

🧠 Core Research Projects (PhD)

🧠 ShapBPT — Explainable AI for Computer Vision

PDF arXiv Code Tests PyPI User Study Poster

• Hierarchical Shapley-based image explanations
• Data-aware Binary Partition Trees (BPT)
• Improved localization + explanation stability

📊 LIME Stratified Sampling

Paper Code Tests PyPI Slides

• Improved LIME sampling strategy
• Reduced variance in perturbations
• More stable and reliable explanations

🔍 Explainable Anomaly Detection (XAD)

Code

• Combined VAE–GAN + XAI
• Studied trust in anomaly detection
• Human-centered interpretability insights

⚙️ Systems & Deployment

⚡ AI on Edge Devices

Code

• Lightweight models for edge deployment
• Model compression & quantization
• Deployment on Raspberry Pi

🏭 Applied / Freelance Research (2017–2022)

Worked on international academic & industrial projects in:

🔬 Domains

• Medical Imaging (MRI, CT, dermoscopy, fundus)
• Computer Vision pipelines (segmentation, enhancement, stitching)
• Intelligent decision systems (ML, ensembles, optimization)

🛠️ Technologies

MATLAB, Python, Deep Learning, Machine Learning

📦 Deliverables

• End-to-end systems with GUI
• Solutions for healthcare and safety-critical applications

🔗 Repository:
https://github.com/rashidrao-pk/r4sshd/tree/main/project_completed

💬 Client Feedbacks:
https://github.com/rashidrao-pk/rashidrao-pk/blob/main/project_completed/feedbacks

🧠 Summary

• Explainable AI + Anomaly Detection (core focus)
• Industrial AI systems (EU projects)
• Research → Deployment pipelines
• Real-time and interpretable AI systems