Alexander Marson M

Key Skills

Software

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Video

Top Task Types

Bounding Box

Point Key Point

Segmentation

Classification

Tracking

Emotion Recognition

Freelancer Overview

I am an AI Training Specialist with 3 years of hands-on experience in data labeling and annotation for image, video, and audio datasets. My work has supported the development of computer vision, natural language processing, and speech recognition models, where I have performed tasks such as bounding box, polygon, and keypoint annotation, as well as semantic segmentation and multi-object tracking. I am skilled in using annotation tools like Labelbox, CVAT, Supervisely, and VIA, and have prepared datasets in YOLO, COCO, and Pascal VOC formats for object detection models. My experience also includes audio transcription, sentiment and intent classification, and rigorous QA to ensure high-quality, consistent datasets. I enjoy collaborating with machine learning engineers to refine annotation guidelines and optimize workflow management for diverse AI projects.

Labeling Experience

Autonomous Vehicle Object Detection & Tracking Dataset Annotation

LabelboxVideoBounding BoxPoint Key Point

Currently working on large-scale video annotation projects focused on computer vision model training and performance optimization. The project involves detailed frame-by-frame labeling of dynamic environments to support object detection, tracking, and action recognition systems. Key responsibilities include: Performing frame-by-frame object detection using bounding boxes and polygon segmentation Executing multi-object tracking across consecutive video frames Annotating moving objects such as vehicles, pedestrians, cyclists, and other dynamic entities Labeling actions and behaviors for action recognition models Handling occlusions, motion blur, and complex real-world scenarios Ensuring annotation consistency according to strict labeling guidelines Conducting quality assurance checks to maintain high dataset accuracy Preparing datasets in YOLO and COCO formats for model training The project involves high-volume video datasets captured in real-world environments, requiring preci

Currently working on large-scale video annotation projects focused on computer vision model training and performance optimization. The project involves detailed frame-by-frame labeling of dynamic environments to support object detection, tracking, and action recognition systems. Key responsibilities include: Performing frame-by-frame object detection using bounding boxes and polygon segmentation Executing multi-object tracking across consecutive video frames Annotating moving objects such as vehicles, pedestrians, cyclists, and other dynamic entities Labeling actions and behaviors for action recognition models Handling occlusions, motion blur, and complex real-world scenarios Ensuring annotation consistency according to strict labeling guidelines Conducting quality assurance checks to maintain high dataset accuracy Preparing datasets in YOLO and COCO formats for model training The project involves high-volume video datasets captured in real-world environments, requiring preci

2024

High-precision Multimodal Data Annotation for computer Vission Models

LabelboxImageBounding BoxPoint Key Point

Worked on large-scale annotation of image and video datasets for autonomous vehicle perception systems. The project involved frame-by-frame object detection and multi-object tracking across urban and highway driving scenarios. Responsibilities included: Annotating vehicles, pedestrians, cyclists, traffic signs, and road infrastructure using bounding boxes and polygon segmentation Performing multi-object tracking across consecutive video frames Creating YOLO-formatted datasets for training object detection models Labeling complex scenarios such as occlusions, night driving, and crowded intersections Following strict annotation guidelines to ensure dataset consistency Conducting quality assurance reviews and correcting annotation errors Collaborating with machine learning engineers to refine class definitions and edge-case handling The dataset included over 150,000 annotated images and 2,000+ video sequences. Maintained 98%+ QA accuracy rate across submissions.

Worked on large-scale annotation of image and video datasets for autonomous vehicle perception systems. The project involved frame-by-frame object detection and multi-object tracking across urban and highway driving scenarios. Responsibilities included: Annotating vehicles, pedestrians, cyclists, traffic signs, and road infrastructure using bounding boxes and polygon segmentation Performing multi-object tracking across consecutive video frames Creating YOLO-formatted datasets for training object detection models Labeling complex scenarios such as occlusions, night driving, and crowded intersections Following strict annotation guidelines to ensure dataset consistency Conducting quality assurance reviews and correcting annotation errors Collaborating with machine learning engineers to refine class definitions and edge-case handling The dataset included over 150,000 annotated images and 2,000+ video sequences. Maintained 98%+ QA accuracy rate across submissions.

2021 - 2024

Education

Work History