Phillip Downs

Key Skills

Software

CVAT

Data Annotation Tech

iMerit

Mindrift

OneForma

OpenCV AI Kit (OAK)

Remotasks

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Document

Geospatial Tiled Imagery

Image

Medical Dicom

Video

Top Label Types

Bounding Box

Computer Programming Coding

Entity Ner Classification

Object Detection

Tracking

Translation Localization

Freelancer Overview

Dedicated Data Annotation Specialist with a core focus on high-fidelity video tracking and temporal analysis. I specialize in multi-object tracking (MOT), frame-by-frame interpolation, and complex action recognition across diverse datasets. My expertise lies in maintaining strict spatial and temporal consistency, ensuring that unique identifiers remain accurate even during heavy occlusions or rapid environmental changes. Beyond basic bounding boxes, I am proficient in semantic segmentation and key-point labeling for human pose estimation. I pride myself on a 98%+ QA pass rate and a deep understanding of how edge cases in labeling—such as truncated objects or motion blur—impact model training. I am highly adaptable to proprietary tooling and have a proven track record of delivering clean, structured data for autonomous systems and surveillance AI.

Labeling Experience

Multi-Object Tracking & Temporal Action Annotation for Autonomous Systems

CVATVideoBounding BoxEntity Ner Classification

Focused on high-fidelity video annotation to train computer vision models for dynamic environments. My work involves frame-by-frame object tracking and interpolation, ensuring that unique IDs remain consistent across occlusions and lighting changes. I specialize in identifying complex interactions, such as pedestrian intent and vehicle maneuvers, while maintaining a low Error Rate in attribute classification (e.g., distinguishing between emergency vehicles and standard transport).

Focused on high-fidelity video annotation to train computer vision models for dynamic environments. My work involves frame-by-frame object tracking and interpolation, ensuring that unique IDs remain consistent across occlusions and lighting changes. I specialize in identifying complex interactions, such as pedestrian intent and vehicle maneuvers, while maintaining a low Error Rate in attribute classification (e.g., distinguishing between emergency vehicles and standard transport).

2019 - 2024

Education

Work History