Dawit Tsegaye

Key Skills

Software

CVAT

Top Subject Matter

No subject matter listed

Top Data Types

Image

Top Label Types

Bounding Box

Segmentation

Freelancer Overview

I am a data annotation and AI training specialist with hands-on experience in high-precision computer vision labeling, particularly for object detection tasks using Roboflow and YOLO formats. My work includes detailed bounding-box annotation on complex datasets, such as football player detection involving fast motion, occlusions, and crowded scenes. I am skilled in maintaining strict quality control through batch self-review, spot checks, and consistency sweeps, and I am adept at handling edge cases like partial visibility and motion blur. I have a strong technical background in Python and ML workflows, and I understand how annotation quality directly impacts model performance. My experience spans both annotation and broader AI/ML engineering, including NLP projects and recommendation systems, giving me a comprehensive perspective on the importance of high-quality training data in real-world AI applications.

Labeling Experience

data annotation

CVATImageBounding BoxSegmentation

I worked on a large-scale image data labeling project focused on football players and football-related scenarios to support computer vision and object detection models. The work involved accurately annotating images by drawing bounding boxes around football players and key objects such as the ball and referees. The dataset included images from matches and training sessions with varying camera angles, lighting conditions, and levels of player occlusion. I followed strict annotation guidelines to ensure labeling consistency and high accuracy across the dataset. Quality control was a key part of the project, including reviewing annotations, correcting errors based on feedback, and meeting defined accuracy and productivity standards. The labeled data contributed to improving the performance of AI models used for sports analytics and automated visual recognition.

I worked on a large-scale image data labeling project focused on football players and football-related scenarios to support computer vision and object detection models. The work involved accurately annotating images by drawing bounding boxes around football players and key objects such as the ball and referees. The dataset included images from matches and training sessions with varying camera angles, lighting conditions, and levels of player occlusion. I followed strict annotation guidelines to ensure labeling consistency and high accuracy across the dataset. Quality control was a key part of the project, including reviewing annotations, correcting errors based on feedback, and meeting defined accuracy and productivity standards. The labeled data contributed to improving the performance of AI models used for sports analytics and automated visual recognition.

2021 - 2023

Education

Work History