Adwait T.

Key Skills

Software

LabelImg

Roboflow

CVAT

Label Studio

Top Subject Matter

Computer Vision data preparation for autonomous traffic monitoring (object detection)

Computer Vision segmentation dataset preparation (polygon annotation)

NLP labeling for sentiment analysis and NER tagging

Top Data Types

Image

Text

Top Task Types

Bounding Box

Polygon

Entity (NER) Classification

Freelancer Overview

Object Detection dataset annotation (Bounding Boxes). Core strengths include LabelImg, Roboflow, and CVAT. AI-training focus includes data types such as Image and Text and labeling workflows including Bounding Box, Polygon, and Entity (NER) Classification.

Labeling Experience

Text classification (sentiment) and NER tagging projects

Label StudioTextEntity (NER) Classification

Conducted text labeling for sentiment classification and NER tagging. Labeled e-commerce client reviews and social media comments into multi-tier sentiment categories. Extracted and cataloged entities such as locations, dates, and brand names to enrich structured training data. • Categorized reviews/comments as Positive, Negative, or Neutral. • Applied Named Entity Recognition (NER) to identify key variables. • Labeled multiple entity types including locations, dates, and brand names. • Used consistent tagging conventions to support downstream NLP tasks.

Present

Semantic segmentation & polygon annotation using CVAT

CVATImagePolygon

Performed semantic segmentation and polygon mapping using CVAT for complex imagery. Created detailed polygon annotations for irregular shapes and environmental landmarks. Resolved overlap and refined edges to maintain consistent, high-quality labels. • Annotated highly complex geometries including aerial imagery and irregular objects. • Addressed overlapping attributes to reduce label ambiguity. • Improved edge identification for more accurate segmentation masks. • Produced consistent ground truth for segmentation model training.

Present

Object Detection dataset annotation (Bounding Boxes)

LabelImgImageBounding Box

Designed and annotated an object detection dataset to support autonomous traffic monitoring. Used bounding box labeling to accurately identify key classes such as vehicles, pedestrians, and traffic signals. Ensured high-fidelity ground truth by following strict pixel-perfect boundary guidelines. • Labeled 500+ images for traffic monitoring scenarios. • Applied bounding boxes with careful attention to boundary precision. • Minimized background noise to improve downstream ML training quality. • Followed dataset guidelines and performed quality-oriented verification.

Present