georaiser
Jorge Rodríguez
Professional Profile
Jorge Rodríguez is a Geospatial Data Scientist, Full Stack Java Developer, and AI Enthusiast specializing in Geomatics, Remote Sensing, and AI-driven solutions. He combines expertise in spatial data analysis with cutting-edge artificial intelligence and enterprise software development.
Core Competencies
Technical Skills
- Programming: Java, Python, HTML/CSS/JavaScript
- AI/ML Frameworks: PyTorch, TensorFlow, OpenCV
- Backend Development: Spring Boot, RESTful APIs, Microservices
- Geospatial: QGIS, ArcGIS, GDAL, PostGIS
- Specializations: Deep Learning, Computer Vision, Object Detection, Remote Sensing
Featured Projects
1. Object Detection Benchmark
Purpose: Comprehensive performance comparison of state-of-the-art object detection models (YOLO v11, Faster R-CNN, RetinaNet, SSD)
Description: This project presents an in-depth comparative analysis of multiple state-of-the-art object detection models, evaluating their performance across diverse metrics including accuracy, processing speed, and computational resource utilization. The benchmark provides data-driven insights for selecting the most appropriate detection model based on specific application requirements and constraints.
Technologies: Python, PyTorch, OpenCV, YOLO v11
Key Achievement: Created reproducible benchmarking framework analyzing speed-accuracy trade-offs for model selection in production environments
Learning: Understood that model selection is context-dependent—YOLO excels in real-time scenarios (60+ FPS) while Faster R-CNN provides superior accuracy when processing time is less critical.
2. Real-Time Object Tracking System
Purpose: Multi-object tracking combining Faster R-CNN detection with DeepSORT algorithm, featuring intelligent counting with virtual line crossing detection
Description: An advanced multi-object tracking system that seamlessly integrates Faster R-CNN for precise object detection with DeepSORT for robust tracking across video frames. The system features intelligent counting mechanisms with configurable virtual line crossing detection, making it applicable to traffic monitoring, crowd analysis, retail analytics, and surveillance applications.
Technologies: Faster R-CNN, DeepSORT, Python, PyTorch, OpenCV, Kalman Filters
Key Achievement: Developed robust tracking system handling occlusions and maintaining persistent object IDs across video frames, suitable for traffic monitoring and surveillance applications
Learning: Mastered the complexity of data association problems, understanding how appearance features (CNN embeddings) and motion prediction (Kalman filters) work together for reliable tracking.
3. Instance Segmentation with Mask R-CNN
Purpose: Pixel-level object segmentation using Mask R-CNN on Pascal VOC dataset, advancing from bounding box detection to precise instance masks
Description: A comprehensive implementation showcasing both object detection and pixel-precise instance segmentation using Faster R-CNN and Mask R-CNN architectures. Trained on the Pascal VOC dataset, this project demonstrates the evolution from simple bounding box detection to sophisticated pixel-level segmentation, enabling more precise object localization crucial for advanced computer vision applications in robotics, autonomous systems, and medical imaging.
Technologies: Mask R-CNN, Faster R-CNN, PyTorch, Pascal VOC Dataset, OpenCV
Key Achievement: Implemented dual-mode system comparing detection vs. segmentation, achieving >0.7 IoU for high-quality instance masks across 20 object categories
Learning: Discovered that instance segmentation requires 2-3x more computational resources than detection but provides invaluable precision for applications like robotics, medical imaging, and autonomous systems. Transfer learning from COCO pre-trained weights proved essential for faster convergence.
Professional Objectives
- Innovate at the intersection of AI, geospatial technology, and software engineering
- Develop scalable solutions addressing real-world challenges in spatial data analysis
- Contribute to cutting-edge research in computer vision and remote sensing
- Build enterprise applications leveraging geospatial intelligence
Additional Experience
- Geospatial Analysis: Satellite image classification, change detection, spatial data pipelines
- Full Stack Development: RESTful geospatial services, microservices architecture, GIS web applications
- Research: Active engagement with computer vision conferences (CVPR, ICCV), open-source contributions
Object Detection Benchmark
Comparative analysis. |
Faster R-CNN and DeepSORT
Faster R-CNN and DeepSORT. |
Object Detection Benchmark
Comparative analysis. |
Object Detection - Instance Segmentation
object detection and instance segmentation using Faster R-CNN and Mask R-CNN on Pascal VOC dataset. |
Get in Touch
