Hezron L.

Key Skills

Software

Appen

Clickworker

Axiom AI

Argilla

Top Subject Matter

Geospatial & Remote Sensing Data

Disaster Response & Infrastructure Risk

ESG, Sustainability & Regulatory Compliance

Top Data Types

Image

Video

Text

Top Task Types

Segmentation

Object Detection

Text Generation

Classification

Question Answering

Text Summarization

Transcription

Data Collection

Polygon

Freelancer Overview

Image annotation contributor with experience in environmental monitoring and post-disaster infrastructure damage datasets for computer vision model training. Specialized in polygon segmentation, multi-class classification, and change detection across satellite and aerial imagery. Contributed to structured annotation initiatives involving land cover classification, deforestation detection, shoreline change monitoring, flood boundary mapping, and disaster-related structural damage assessment. Worked within standardized taxonomies and detailed annotation guidelines to ensure spatial precision and cross-image consistency. Experienced in multi-temporal image comparison, environmental degradation differentiation, and damage severity classification using structured JSON and GeoJSON outputs compatible with GIS and machine learning pipelines. Participated in blind review cycles and consistency validation processes to maintain high annotation accuracy and reliability in risk-sensitive domains.

Labeling Experience

Post-Disaster Infrastructure Damage Image Annotation

ImageSegmentation

Contributed to a structured image annotation initiative supporting computer vision models designed to assess infrastructure damage following natural disasters. Annotated aerial and satellite imagery across hurricane, flood, and storm-affected regions to improve AI-based damage classification systems used in disaster response and risk modeling. Primary responsibilities included: • Polygon segmentation of damaged residential and commercial structures • Bounding box annotation of partially collapsed buildings • Road obstruction detection (debris, flooding, structural failure) • Flood boundary marking across urban and semi-urban zones • Multi-class damage severity classification (No Damage / Minor / Moderate / Severe / Destroyed) Worked within standardized annotation guidelines to ensure consistent damage severity interpretation across image sets. Participated in: • Blind quality review cycles • Damage severity consistency scoring • Edge-case classification handling (roof discoloration vs structural damage, temporary flooding vs permanent damage) • Structured JSON output formatting compatible with computer vision training pipelines Demonstrated strong visual discrimination skills and consistency in high-variability post-disaster imagery.

2025 - 2025

Environmental Risk & Land Degradation Satellite Annotation Dataset

ImageSegmentation

Contributed to a structured satellite image annotation initiative supporting computer vision models for environmental monitoring and land-use change detection. Annotated multi-temporal satellite imagery across forested, agricultural, coastal, and semi-urban regions to improve land cover classification and environmental risk detection accuracy. Primary responsibilities included: • Polygon segmentation of deforestation zones and degraded land areas • Multi-class land cover classification across a 30+ category environmental taxonomy • Shoreline boundary marking and coastal change detection • Flood-prone region tagging • Differentiation of seasonal vegetation cycles from permanent environmental degradation Worked within established annotation guidelines to ensure spatial precision and cross-image consistency. Participated in: • Blind re-annotation review cycles • Consistency scoring validation (0.85+ agreement range) • Edge-case handling for mixed land-use and transitional zones • Structured JSON / GeoJSON labeling outputs compatible with GIS pipelines Demonstrated strong spatial reasoning, environmental interpretation accuracy, and disciplined adherence to structured annotation standards.

2024 - 2024

Urban Zoning & Infrastructure Compliance Annotation Dataset

ImageClassification

Designed and executed an image-based annotation project focused on training AI systems to interpret urban zoning maps and infrastructure layouts for regulatory compliance and land-use classification. Annotated 9,800 high-resolution zoning maps, satellite overlays, and municipal planning diagrams across 6 urban jurisdictions. The dataset supported AI training for automated land-use detection, development constraint identification, and infrastructure proximity analysis. Annotation tasks included: • Polygon segmentation of zoning boundaries (residential, commercial, mixed-use, industrial, environmental overlays) • Bounding box labeling of infrastructure elements (roads, utilities, public facilities, transit lines) • Pixel-level segmentation of restricted zones (floodplains, heritage districts, setback areas) • Multi-class land-use classification • Conflict-zone tagging where overlapping regulatory layers existed Developed a structured taxonomy of 42 zoning and infrastructure categories to ensure cross-jurisdiction consistency. Quality assurance measures: • 15% blind re-annotation sampling • Inter-annotator agreement validation (0.84 consistency score) • Edge-case stress testing for mixed-use and overlay zones • Standardized annotation guidelines documentation (28-page rulebook) Dataset was delivered in structured JSON format compatible with GIS and computer vision model pipelines. Post-training evaluation results (internal testing): • 29% reduction in zoning misclassification • 34% improvement in regulatory boundary detection • 21% increase in infrastructure proximity accuracy

2024 - 2024

Education

Work History