Adeoye Adewale

Key Skills

Software

Remotasks

Roboflow

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Image

Video

Top Label Types

Action Recognition

Bounding Box

Object Detection

Polygon

Freelancer Overview

With a specialized focus on Computer Vision and Machine Learning dataset labelling, I have gained extensive experience in data annotation through freelance roles with Remotask and Atlas Capture. My technical proficiency includes high-precision image and video labeling using bounding boxes, polygons, and attribute tagging for diverse applications, ranging from autonomous driving to human egocentric analysis. Uniquely, I bridge the gap between annotator and developer. My final year university project involved not just data curation but the actual development of an Intelligent Transport System using Roboflow algorithms. This experience provided me with a critical understanding of model training requirements. I bring a detail oriented, developer 'aware' perspective to data annotation workflows that ensures high quality training data for robust AI models.

Labeling Experience

Video Data Annotator

CrowdsourceVideoAction Recognition

I specialize in Human Egocentric Video Understanding, contributing to the development of AI models for wearable technology and augmented reality systems. My core responsibility involves rigorous segment of frame analysis to perform segmentation, identifying precise start and end timestamps for fine grained human actions and object interactions. This work requires a deep focus on temporal details to ensure the AI can accurately recognize complex behaviors from a first-person perspective. To ensure dataset integrity, I adhere to a strict dual-mode annotation schema, applying either "Dense" labeling for rapid, manipulative interactions or "Coarse" labeling for broader spatial movements, while enforcing rigid exclusion rules to prevent mixed styles within a single episode. I consistently maintain high acceptance rates by tracking object consistency and conducting thorough self audits to ensure all actions and segments meet required standard.

I specialize in Human Egocentric Video Understanding, contributing to the development of AI models for wearable technology and augmented reality systems. My core responsibility involves rigorous segment of frame analysis to perform segmentation, identifying precise start and end timestamps for fine grained human actions and object interactions. This work requires a deep focus on temporal details to ensure the AI can accurately recognize complex behaviors from a first-person perspective. To ensure dataset integrity, I adhere to a strict dual-mode annotation schema, applying either "Dense" labeling for rapid, manipulative interactions or "Coarse" labeling for broader spatial movements, while enforcing rigid exclusion rules to prevent mixed styles within a single episode. I consistently maintain high acceptance rates by tracking object consistency and conducting thorough self audits to ensure all actions and segments meet required standard.

2025

Image Data Annotator

Scale AIImageBounding Box

I contributed to large scale computer vision datasets focused on Industrial Automation and Autonomous Systems. My core responsibilities involved the high precision labeling of logistics machinery and urban street scenes, utilizing 2D bounding boxes and polygon segmentation to distinguish between static equipment, moving parts, vehicles, and pedestrians. Operating within a fast paced, high volume crowdsourced environment, I processed hundreds of images weekly while adhering to complex, multi-page project guildlines. I consistently maintained a Quality Assurance (QA) score above 90% by rigorously applying client specific guidelines regarding class definitions. Through continuous feedback of the reviewer, I ensured the delivery of pixel-perfect groundtruth data.

I contributed to large scale computer vision datasets focused on Industrial Automation and Autonomous Systems. My core responsibilities involved the high precision labeling of logistics machinery and urban street scenes, utilizing 2D bounding boxes and polygon segmentation to distinguish between static equipment, moving parts, vehicles, and pedestrians. Operating within a fast paced, high volume crowdsourced environment, I processed hundreds of images weekly while adhering to complex, multi-page project guildlines. I consistently maintained a Quality Assurance (QA) score above 90% by rigorously applying client specific guidelines regarding class definitions. Through continuous feedback of the reviewer, I ensured the delivery of pixel-perfect groundtruth data.

2022 - 2022

Education

Work History