top of page

NEURALAB

INTRODUCTION

NEURALAB is an internal research platform at Team Cardinalis designed to accelerate computer vision model development, specifically for object detection. It provides researchers with a visual, code-free interface to build, train, and optimize PyTorch models while leveraging the team's dedicated servers instead of local machines. By simplifying architecture design and hyperparameter tuning, The Lab enables faster experimentation and model comparison, ultimately aiming to identify the most efficient object detection architectures without hardware limitations.

TECHNICAL DIAGRAM

Client side

Blueprint

3D Render

JSON Build

Server side

PyTorch Model Training

Dataset Checkup

PyTorch Model Build

VERSION 1.0

In the first iteration, NEURALAB featured a 2D neural network designer. This prototype enabled users to manage PyTorch layers through a simple drag-and-drop interface, offering a straightforward, code-free way to build and configure neural architectures.

Image by Mohammad Rahmani
bottom of page