PRIYANSHI SAXENA
Bareilly
Summary
Data Annotation hands-on experience in Computer Vision, OCR correction, and QA validation. Proven track record of annotating over 600,000+ images using tools like CVAT, Labelbox, and improving AI models for traffic, safety, and smart city applications. Strong understanding of YOLOv5, Faster R CNN, and ML data pipelines. Passionate about quality, detail, and enhancing model accuracy through clean, high-quality datasets.
Overview
1
1
year of professional experience
Work History
Data Labelling
07.2024 - 04.2025
- Annotated over 500,000 vehicle number plate images using CVAT with 98–99% accuracy, directly enhancing the performance of license plate recognition models.
- Performed OCR corrections on large-scale vehicle datasets, improving text recognition accuracy by over 20% and reducing model prediction errors.
- Labelled motorcycle riders based on occupancy (single, double, triple) and helmet use, supporting the development of traffic safety models with over 90% classification accuracy.
- Tagged over 100,000 instances of construction safety equipment (helmets, gloves, shoes, suits), helping object detection models achieve 95%+ detection accuracy in workplace safety scenarios.
- Annotated car occupants for seatbelt detection and visibility, improving behaviour recognition models and boosting detection precision by 18%.
- Monitored live dashboards to track model predictions, contributing to a 25% increase in iteration speed during the ML training process.
- Created high-quality ground truth datasets that led to measurable improvements in model training accuracy and reduced data-related errors by 30%.
- Worked in a Linux-based environment to manage large-scale datasets, automate annotation tasks, and streamline cross-team data workflow.
Data Annotator
11.2023 - 04.2024
- Annotated more than 75,000 images for a smart city traffic monitoring project, ensuring accurate object detection for vehicles, road signs, and pedestrians.
- Annotated 120,000 urban scene images to train computer vision models for real-time navigation in autonomous vehicle systems, contributing to safer route planning algorithms.
- Supported the training of object detection models by curating and structuring labelled datasets optimised for YOLOv5 and Faster R-CNN architectures.
- Integrated feedback from data scientists and model trainers, enabling faster dataset refinement and increasing model convergence rates during training.
Education
BBA - Business Administration And Management
Mahatma Jyotiba Phule Rohilkhand University
Bareilly, India 04.2001 -
Master of Arts - Political Science And International Relations
Indira Gandhi National Open University,
Bareilly, India 04.2001 -
Skills
- Programming languages: Python
- Frameworks and Tools: Visual Studio Code
Timeline
Data Labelling
07.2024 - 04.2025
Data Annotator
11.2023 - 04.2024
BBA - Business Administration And Management
Mahatma Jyotiba Phule Rohilkhand University
04.2001 -
Master of Arts - Political Science And International Relations
Indira Gandhi National Open University,
04.2001 -