I'm excited to present my completed Bachelor's thesis, in which I developed a camera-based 3D object detection system for robotic systems. The goal was to combine modern AI methods with efficient edge computing hardware โ for precise and real-time capable solutions in industrial applications.
โข Hardware
โ NVIDIA Jetson Orin Nano (powerful edge device for AI)
โ ZED Mini Stereo Camera (real-time 3D depth sensing)
โข Software & Methods
โ YOLOv8 (deep learning for object detection)
โ PyTorch & CUDA (GPU-accelerated inference)
โ OpenCV (image processing & visualization)
โข Filtering of depth noise for stable 3D data
โข Projection of 3D coordinates onto 2D image planes
โข Real-time calculation of distances and object dimensions
โข Combination of 2D object detection (YOLO) with depth data from the stereo camera
The solution enables precise localization and measurement of objects in real time โ essential for robotic applications in industry, logistics, or autonomous navigation. I also compared various depth estimation methods and optimized performance for real-time deployment.
More details and sample images can be found in the GitHub repository:
๐ GitHub


#AI #Robotics #3DObjectDetection #ComputerVision #EdgeAI #JetsonNano #DeepLearning #3DKamera #YOLO