Jiang Zhenxiang

Key Skills

Software

LabelImg

Top Subject Matter

No subject matter listed

Top Data Types

Video

Top Task Types

Classification

Object Detection

Freelancer Overview

I am passionate about AI and data-driven solutions, with hands-on experience as an AI Trainer where I optimized data pipelines and automated data preparation for over 40 game models, significantly boosting model accuracy and efficiency. My background includes developing and refining machine learning models using Python, TensorFlow, and advanced data processing tools, as well as building computer vision projects such as face recognition systems and reinforcement learning simulations for autonomous vehicles. I am skilled in data annotation, feature engineering, database management, and performance tuning, and I thrive in environments that require meticulous attention to detail and a deep understanding of AI-driven workflows. My experience spans computer vision, image processing, and predictive modeling, and I am eager to contribute my expertise to projects focused on high-quality AI training data.

Labeling Experience

Game annotations

LabelimgVideoClassificationObject Detection

This project builds a game UI-based event and state recognition system that leverages OCR and other computer vision techniques to automatically detect key in-game events such as deaths and level-ups from live or recorded gameplay footage. It captures frames from the game screen, locates and reads UI elements (e.g., death screens, level-up prompts), and then classifies them into predefined event types to support automated logging, statistics, and potential downstream applications like highlight generation or gameplay analysis.

This project builds a game UI-based event and state recognition system that leverages OCR and other computer vision techniques to automatically detect key in-game events such as deaths and level-ups from live or recorded gameplay footage. It captures frames from the game screen, locates and reads UI elements (e.g., death screens, level-up prompts), and then classifies them into predefined event types to support automated logging, statistics, and potential downstream applications like highlight generation or gameplay analysis.

2024 - 2024

Education

Work History