Finished my BSc Thesis!
🚀 Thrilled to Announce My Completed BSc Thesis! 🚀
I’m excited to share that I’ve finished my BSc thesis, Unsupervised Camera-LiDAR Fusion for Fast Region Proposal! After months of intense research, development, and analysis, this project has been a journey of merging advanced sensor technologies to tackle real-world challenges in object detection. 🌍🤖
📜 Thesis Details
The thesis, written in Farsi, is now available for those interested in delving deeper! You can check it out here, and if you want a quick overview, my presentation slides are accessible here. My work was supervised by the knowledgeable Mahdi Javanmardi, with Mohammad Rahmati as a committee member—two mentors whose guidance was invaluable throughout this journey.
Thesis Abstract:
🎯 Why This Research?
In recent years, the integration of multiple sensory data sources—like cameras and LiDAR—has revolutionized object detection systems, bringing an edge in both accuracy and speed. However, the process of collecting and manually labeling training data for these models can be incredibly costly and time-consuming.
🤖 What Did We Do?
In this thesis, I proposed an unsupervised learning approach to fuse camera and LiDAR data for region proposal, focusing on how this method can impact the effectiveness, accuracy, and speed of object detection. By blending 2D camera images with 3D LiDAR data, and using unsupervised clustering algorithms, we identified potential object regions without the need for labeled data—a significant time-saver in the development pipeline.
📊 What Did We Find?
The results showed that our approach could identify object-containing regions with a high degree of accuracy, even without labeled data. This means that this method has the potential to reduce both the cost and time required to develop high-performing object detection systems—paving the way for faster and more efficient applications in autonomous vehicles, surveillance, and beyond!
Thank you to everyone who supported me on this journey, and to all those fascinated by the future of sensor fusion and AI—there’s so much more to explore! 🌌
Stay tuned for more updates, and feel free to dive into the full thesis and slides if you’re interested in the technical details!