About
Lightweight-AI is a next-generation deep learning platform designed for speed, portability, and real-world deployment—especially in medical imaging. Our framework enables high-accuracy, low-latency AI even on resource-constrained hardware, as demonstrated in our recent publications:
npj Digital Medicine (2025): “A hybrid multi-model artificial intelligence approach for glaucoma screening using fundus images”.
Scientific Reports (2022): “A lightweight deep learning model for automatic segmentation and analysis of ophthalmic images”.
Key Features
- Ultra-efficient lightweight neural networks for edge & cloud
- Open-source Python API: PyPI: Lightweight-AI
- Patent granted on core lightweight medical imaging methods*
- Supports real-time screening on standard laptops, Raspberry Pi, and embedded devices
- Validated on large-scale multicenter ophthalmic datasets
Glaucoma, a leading cause of blindness, requires accurate early detection. We have developed an AI-based Glaucoma Screening (AI-GS) network with lightweight deep learning models that analyze fundus images to identify early structural signs such as optic disc cupping, hemorrhages, and nerve fiber layer defects. By making accurate glaucoma screening accessible at convenience, the AI-GS network offers a scalable solution to prevent progression of glaucoma and vision loss.
AIGS Network: End-to-End Screening Workflow
Detection, diagnosis, and treatment of ophthalmic diseases depend on extracting information (features or dimensions) from images. Deep learning (DL) models are crucial for automating this process. We designed a DL model (LWBNA-Unet), ~10× lighter than the standard UNet for biomedical image segmentation. Our models can be trained to segment a variety of ophthalmic images precisely and classify diseases.
Lightweight Deep Learning Model Architecture
Title: Biological image processing program, biological image processing device, and biological image processing method
Patent No: WO2023062764A1
Inventors: Toru Nakazawa & Parmanand Sharma
Year: 2023
Download Patent PDF
Members
- Team list coming soon.
Get Started:
- Read our papers: npj Digital Medicine, Scientific Reports
- Install via pip:
pip install Lightweight-AI - Project source: PyPI / Documentation
This is a research demo; not a medical device. Uploaded images may be retained to improve robustness—please remove any personal info from filenames/metadata.Use is limited to lawful, non-clinical research and education; no warranty is provided; availability may change without notice.