Ahmed Ghida

computer vision freelancer

• Job Details:As a computer vision freelancer, I have delivered multiple ai-driven solutions focused on document automation, industrial safety, and visual inspection. I built an end-to-end image classification pipeline using python and tensorflow to process academic pdfs, including scraping data from arxiv, converting pages into images, and deploying a model for page type and orientation detection. I also developed a safety compliance system for construction sites using pose estimation and object detection to monitor ppe usage and worker counts, complete with a streamlit dashboard and docker-based deployment. In another project, I implemented a crack segmentation system that converted pixel-level masks into real-world area measurements (in mm²), exported results to csv, and generated visual summaries. I also created an ocr pipeline tailored for extracting structured data from german-language invoices. Additionally, I fine-tuned yolov8 for icon detection in small datasets, used resnet50 for feature extraction, and developed a similarity-matching function to identify icons, supported by ocr as a fallback. These projects demonstrate my ability to design, build, and deploy scalable, real-world computer vision solutions.