Data Synthesis

Another unique strength of Pebblous is its high-quality synthetic data. When data is scarce or expensive to obtain, using synthetic data to train AI is the most powerful and efficient approach.

Christmas Capsules, 2019. Digital media, Lee Ju-haeng

Generate precision targeting synthetic data based on quality diagnosis

Computer Graphics

Use advanced CG software and rendering technology to create visually realistic images or animations.

Generative

Using a deep learning model, it automatically generates new data that mimics the real world data patterns.

Modeling & Simulation

After creating a virtual environment based on scenarios and equations that express the real world, we create data by modeling and simulating the process and behavior of the real world.

Synthetic Dataset: Case Study

Single image 3D creation

Single photo -based 3D model generation synthetic data

* POSTECH joint research, achievement of SOTA levels(SIGGRAPH Asia 2023)

Drone Detection

3D synthesis data for drone awareness and tracking

Characters and poses in special environments

Synthesis data for AI learning for figures and attitude recognition in a special environment, such as children's perception within the vehicle

Robot autonomous driving field

This is a dataset required for robots such as vacuum cleaners, and AI detects objects and plans optimal driving paths.

animal behavior

Synthesis data for AI learning to classify mouse behavior and recognize body parts

Livestock sector

After creating a stable diffusion -based LORA model for analysis of small behavioral forms

Diet monitoring

Synthetic data for meals for monitoring. Use hybrids of 3D and generated models

Logistics field

This is a dataset for automatically recognizing the inventory of the stand in an environment such as an unmanned shop.

Pharmaceutical field

In the automatic pill automatic preparation system of large hospitals, a computer vision inspection requires a pill dataset of various forms and configuration. Synthesis data for the detection and coefficient of the pill

Waste Plastic Recycling Classification AI Synthetic Data

Synthetic data for AI learning used to recycle resources through plastic material classification