John Mark

Key Skills

Software

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Video

Audio

Top Label Types

Bounding Box

Point Key Point

Segmentation

Classification

Tracking

Emotion Recognition

Translation Localization

Evaluation Rating

Transcription

Freelancer Overview

I am an AI training specialist with over three years of hands-on experience in data annotation and labeling for machine learning and computer vision projects. My expertise spans image, video, and audio annotation, including bounding boxes, polygons, segmentation, multi-object tracking, and timestamped transcription for NLP and speech recognition models. I have a strong track record of preparing high-quality, YOLO-compatible datasets and ensuring 99%+ annotation accuracy through rigorous quality assurance. I am proficient with tools such as Labelbox, CVAT, Roboflow, Supervisely, VIA, and Amazon SageMaker Ground Truth, and have supported large-scale projects like autonomous vehicle object detection and multilingual speech recognition. I am passionate about delivering accurate, scalable training data and collaborating with AI teams to drive model performance.

Labeling Experience

Autonomous Object Detection & Multi-Object Tracking Dataset

LabelboxVideoBounding BoxPoint Key Point

I am currently working on a large-scale video annotation project focused on training and improving computer vision models for real-time object detection, action recognition, and multi-object tracking. The project involves annotating high-resolution video footage across diverse environments, including urban outdoor scenes and complex dynamic settings. My role includes performing detailed frame-by-frame annotations using bounding boxes and polygon segmentation to accurately label moving objects such as vehicles, pedestrians, and other relevant entities. I assign consistent object IDs across sequential frames to maintain tracking continuity and ensure temporal accuracy throughout the video sequences. In addition, I annotate human actions and interactions to support behavior recognition models. I handle challenging scenarios such as occlusions, motion blur, overlapping objects, and low-light conditions while maintaining strict adherence to annotation guidelines. The project has involved

I am currently working on a large-scale video annotation project focused on training and improving computer vision models for real-time object detection, action recognition, and multi-object tracking. The project involves annotating high-resolution video footage across diverse environments, including urban outdoor scenes and complex dynamic settings. My role includes performing detailed frame-by-frame annotations using bounding boxes and polygon segmentation to accurately label moving objects such as vehicles, pedestrians, and other relevant entities. I assign consistent object IDs across sequential frames to maintain tracking continuity and ensure temporal accuracy throughout the video sequences. In addition, I annotate human actions and interactions to support behavior recognition models. I handle challenging scenarios such as occlusions, motion blur, overlapping objects, and low-light conditions while maintaining strict adherence to annotation guidelines. The project has involved

2024

Multilingual Speech Recognition & Audio Transcription Dataset

LabelboxAudioClassificationEmotion Recognition

I worked on a large-scale multilingual audio annotation project designed to improve automatic speech recognition (ASR) and conversational AI systems. The project involved transcribing and labeling thousands of short and long-form audio recordings, including conversational speech, customer service interactions, and real-world background-noise environments. My responsibilities included producing accurate verbatim and clean-read transcriptions with precise timestamp segmentation at sentence and word levels. I annotated speaker differentiation (speaker diarization), labeled emotional tone (e.g., neutral, happy, frustrated), and classified audio clips based on intent and context. I also performed quality evaluation and rating tasks to validate model outputs and improve dataset reliability.

I worked on a large-scale multilingual audio annotation project designed to improve automatic speech recognition (ASR) and conversational AI systems. The project involved transcribing and labeling thousands of short and long-form audio recordings, including conversational speech, customer service interactions, and real-world background-noise environments. My responsibilities included producing accurate verbatim and clean-read transcriptions with precise timestamp segmentation at sentence and word levels. I annotated speaker differentiation (speaker diarization), labeled emotional tone (e.g., neutral, happy, frustrated), and classified audio clips based on intent and context. I also performed quality evaluation and rating tasks to validate model outputs and improve dataset reliability.

2024 - 2024

Autonomous Object Detection & Multi-Object Tracking Dataset

LabelboxImageBounding BoxPoint Key Point

Worked on a large-scale autonomous driving dataset focused on improving object detection and multi-object tracking models. The project involved annotating and tracking vehicles, pedestrians, cyclists, traffic signs, and road markings across 50,000+ video frames and high-resolution images. Key Responsibilities: Created bounding boxes and polygon annotations for dynamic road objects Performed frame-by-frame multi-object tracking using unique IDs Conducted semantic segmentation for lane markings and road boundaries Prepared YOLO-compatible dataset formats for training and validation Maintained 99%+ annotation accuracy through internal QA reviews Followed strict annotation guidelines and edge-case handling procedures Assisted in dataset cleaning, balancing, and validation The dataset was used to train real-time object detection and tracking models for smart mobility systems.

Worked on a large-scale autonomous driving dataset focused on improving object detection and multi-object tracking models. The project involved annotating and tracking vehicles, pedestrians, cyclists, traffic signs, and road markings across 50,000+ video frames and high-resolution images. Key Responsibilities: Created bounding boxes and polygon annotations for dynamic road objects Performed frame-by-frame multi-object tracking using unique IDs Conducted semantic segmentation for lane markings and road boundaries Prepared YOLO-compatible dataset formats for training and validation Maintained 99%+ annotation accuracy through internal QA reviews Followed strict annotation guidelines and edge-case handling procedures Assisted in dataset cleaning, balancing, and validation The dataset was used to train real-time object detection and tracking models for smart mobility systems.

2021 - 2024

Education

Work History