AI Technologies by MSISLAB
We are developing advanced deep learning technologies
with self-training and light-weight Edge AI
for Smart Factory and Autonomous Driving.
Motion Anomaly Detection with Self-Training Deep Learning
Product Defect Detection with Self-Reinforcement Deep Learning
Self-Supervised training with Self Data Augmentation Requiring no labeled dataset
Self-Training AI Technology
MSISLAB's Breakthrough Unsupervised Learning & Self-Supervised Learning AI Technology
Unsupervised Learning using only normal video data
Teaching itself with Zero anomaly data, zero labeling
Only 20 min training provides accuracy up to 99.99%
Self-Training AI Robot to understand manufacturing process based on the product type and shape
Fully automate the process of assembly, inspection, packaging with no humans in the factory lines
Multi-Camera based Vehicle Tracking Using Collaborative Deep Learning
Self-Supervised Deep Learning for Vehicle Tracking and and Pedestrian Action Recognition
Compact CNN Accelerator SoC for AI Robot
Scalable Multi-Chip Architecture to support the fast evolution of deep learning AI models
Power Optimized for mobile AI robot vision on battery