Mohammed F.

Key Skills

Software

Other

Telus

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Image

Video

Top Task Types

Tracking

Freelancer Overview

With hands-on experience in high-precision video annotation, I contributed to the Poland Soccer Project by annotating jersey numbers and tracking player identities using RespoAnnotationApp (Autocom, OCR, All mode). I verified algorithmic outputs and manually identified players through visual features such as hairstyle, facial hair, sleeve length, cleats, tattoos, and field positioning. Working in multi-user workflows with GCP sync, I delivered structured files (, , ) and achieved over 98% QA accuracy, strengthening expertise in sports analytics and AI training data. Building on this, I worked on Project Tiramisu for self-driving cars, annotating lane lines, road boundaries, centerlines, arrows, and traffic signs using GT Studio, 2D, and 3D views. I applied attributes including color, type (solid/dashed), visibility (camera/lidar), and lane usage (HOV, bus/taxi, turn lanes). I validated 3D annotations to ensure lane lines touched the ground correctly without zigzag or floating errors, producing structured outputs with high QA accuracy. Skilled in 2D/3D annotation, attribute management, and quality assurance, I deliver clean, reliable datasets for autonomous driving AI.

Labeling Experience

Self‑Driving Car Annotation – Project Tiramisu

Telus3D SensorCuboidClassification

Annotated lane lines, road boundaries, centerlines, arrows, and traffic signs using telus GT Studio, 2D, and 3D views. Applied attributes such as color, type (solid/dashed), visibility (camera/lidar), and lane usage (HOV, bus/taxi, turn lanes). Verified 3D annotations to ensure lane lines touched the ground correctly without zigzag or floating errors. Delivered structured outputs with high QA accuracy.

2025

Soccer Video Annotation – Poland Project

OtherVideoClassificationTracking

Annotated jersey numbers and tracked player identities using RespoAnnotationApp (Autocom, OCR, All mode). Verified outputs and manually identified players through visual features such as hairstyle, facial hair, sleeve length, cleats, tattoos, and field positioning. Delivered structured files (identity.csv, timings.csv, track_annotations.csv) in multi-user workflows with GCP sync. Achieved >98% QA accuracy.

2025

Education

Work History