English

SAKURA-I Model Library for Edge AI Applications

MERA developers get a head start with our Model Zoo, pre-trained AI inference models optimized for the SAKURA AI Accelerator powered by DNA. Code drops into MERA, ready to run or modify. Applications include classification, object detection, segmentation, pose estimation, and more.

Featured AI Inference Models

lidar
SFA3D
Framework: PyTorch
3D LiDAR Object Detection
Resolution: 608x608
Get the Model
vision
MonoDepth
Framework: PyTorch
Monocular Depth Estimation
Resolution: 384x288
Get the Model
pose-estimation
MoveNet Thunder
Framework: TFLite
Pose Estimation
Resolution: 256x256
Get the Model
Filter by:
Clear Filters
Model
ResNet18-v1.5
Framework
PyTorch
Application
Classification
Input Resolution
224x224
Calibration Data
Real-Data
Get the Model
Model
ResNet50-v1.5
Framework
PyTorch
Application
Classification
Input Resolution
224x224
Calibration Data
Real-Data
Get the Model
Model
YoloV3
Framework
TFLite
Application
Object-Detection
Input Resolution
416x416
Calibration Data
Real-Data
Get the Model
Model
Yolov5s
Framework
TFLite
Application
Object-Detection
Input Resolution
448x448
Calibration Data
Real-Data
Get the Model
Model
YoloV5m
Framework
TFLite
Application
Object-Detection
Input Resolution
640x640
Calibration Data
Real-Data
Get the Model
Model
SFA3D
Framework
PyTorch
Application
3D-LiDAR-Object-Detection
Input Resolution
608x608
Calibration Data
Real-Data
Get the Model
Model
EfficientNet-Lite-0
Framework
TFLite
Application
Classification
Input Resolution
240x240
Calibration Data
Real-Data
Get the Model
Model
EfficientNet-Lite-2
Framework
TFLite
Application
Classification
Input Resolution
260x260
Calibration Data
Real-Data
Get the Model
Model
EfficientNet-Lite-3
Framework
TFLite
Application
Classification
Input Resolution
280x280
Calibration Data
Real-Data
Get the Model
Model
EfficientNet-Lite-4
Framework
TFLite
Application
Classification
Input Resolution
300x300
Calibration Data
Real-Data
Get the Model
Model
EfficientNetV2-B0
Framework
TFLite
Application
Classification
Input Resolution
224x224
Calibration Data
Random-Data
Get the Model
Model
EfficientNetV2-B1
Framework
TFLite
Application
Classification
Input Resolution
224x224
Calibration Data
Random-Data
Get the Model
Model
EfficientNetV2-B2
Framework
TFLite
Application
Classification
Input Resolution
224x224
Calibration Data
Random-Data
Get the Model
Model
EfficientNetV2-B3
Framework
TFLite
Application
Classification
Input Resolution
224x224
Calibration Data
Random-Data
Get the Model
Model
EfficientNetV2-s
Framework
TFLite
Application
Classification
Input Resolution
224x224
Calibration Data
Random-Data
Get the Model
Model
MonoDepth
Framework
PyTorch
Application
Monocular-Depth-Estimation
Input Resolution
384x288
Calibration Data
Real-Data
Get the Model
Model
U-Net
Framework
TFLite
Application
Segmentation
Input Resolution
128x128
Calibration Data
Real-Data
Get the Model
Model
MoveNet-Thunder
Framework
TFLite
Application
Pose-Estimation
Input Resolution
256x256
Calibration Data
Real-Data
Get the Model
Model
YoloV4-Tiny
Framework
TFLite
Application
Object-Detection
Input Resolution
640x640
Calibration Data
Real-Data
Get the Model
Model
DeepLabEdgeTPU-m
Framework
TFLite
Application
Segmentation
Input Resolution
512x512
Calibration Data
Real-Data
Get the Model
Model
DeepLabEdgeTPU-s
Framework
TFLite
Application
Segmentation
Input Resolution
512x512
Calibration Data
Real-Data
Get the Model
Model
MoveNet-Lighting
Framework
TFLite
Application
Pose-Estimation
Input Resolution
192x192
Calibration Data
Real-Data
Get the Model
Model
MobileNetV2-SSD
Framework
PyTorch
Application
Object-Detection
Input Resolution
640x480
Calibration Data
Real-Data
Get the Model
Model
DeepLabEdgeTPU-xs
Framework
TFLite
Application
Segmentation
Input Resolution
512x512
Calibration Data
Real-Data
Get the Model
Model
GladNet
Framework
TFLite
Application
Low-Light-Enhancement
Input Resolution
640x480
Calibration Data
Real-Data
Get the Model
Model
SR-Mobile-Quantization (ABPN)
Framework
TFLite
Application
Super-Resolution
Input Resolution
640x360 to HD
Calibration Data
Real-Data
Get the Model
Model
YoloV7-Quantizer
Framework
MERA
Application
Object-Detection
Input Resolution
640x640
Calibration Data
Real-Data
Get the Model
Model
YoloV4
Framework
TFLite
Application
Object-Detection
Input Resolution
416x416
Calibration Data
Real-Data
Get the Model
Model
SCI-Quantizer
Framework
TFLite
Application
Low-Light-Enhancement
Input Resolution
1280x720
Calibration Data
Real-Data
Get the Model

SAKURA-II M.2 Modules and PCIe Cards

EdgeCortix SAKURA-II can be easily integrated into a host system for software development and AI model inference tasks.

Order an M.2 Module or a PCIe Card and get started today!