Nicola Cappellaro

Key Skills

Software

Roboflow

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Image

Top Label Types

Audio Recording

Bounding Box

Point Key Point

Freelancer Overview

I have hands-on experience in image annotation and AI training data preparation using Roboflow, where I labeled datasets for computer vision tasks including object detection and human pose estimation (skeleton keypoints). My work involved drawing precise bounding boxes, annotating keypoints, and ensuring dataset consistency and quality for machine learning pipelines. I hold a background in Information Engineering, which gives me a strong understanding of how annotated data is used in training AI models. This allows me to produce accurate, consistent labels and quickly adapt to new annotation guidelines and tools. I am detail-oriented, reliable, and comfortable working with large datasets under quality requirements.

Labeling Experience

Computer Vision MoCap

RoboflowImageBounding BoxPoint Key Point

The goal of this project is to estimate the 3D pose of a basketball player recorded using a multi-camera RGB setup. Starting from manually annotated 2D keypoints, the player’s 3D skeleton is reconstructed through geometric triangulation. The estimated poses are then compared with ground-truth MoCap data to assess the accuracy of the triangulation. Additionally, we test a modern 2D human pose estimation algorithm (YOLO-Pose) and evaluate its performance. At the end, we display the 3D skeleton on Unreal Engine.

The goal of this project is to estimate the 3D pose of a basketball player recorded using a multi-camera RGB setup. Starting from manually annotated 2D keypoints, the player’s 3D skeleton is reconstructed through geometric triangulation. The estimated poses are then compared with ground-truth MoCap data to assess the accuracy of the triangulation. Additionally, we test a modern 2D human pose estimation algorithm (YOLO-Pose) and evaluate its performance. At the end, we display the 3D skeleton on Unreal Engine.

2025 - 2025

ESP32 Image Classification

RoboflowImageBounding Box

The goal of this project is to develop a lightweight image classification model that can identify two classes (0 or 1) in grayscale images, and deploy this model on an ESP32 microcontroller.

The goal of this project is to develop a lightweight image classification model that can identify two classes (0 or 1) in grayscale images, and deploy this model on an ESP32 microcontroller.

2024 - 2024

Education

Work History