MobileNet V2 1.0 224 Quant

Model Overview

A quantized MobileNet V2 model with a 1.0 width multiplier optimized for image classification on ImageNet at 224x224 resolution.

The MobileNet V2 1.0 224 Quant model is developed and optimized for the Synaptics Astra™ SL1680 processor NPU and SL1640 processor NPU.

Model Features

• Model Type: Image Classification
• Input Size: 224x224
• Output Size: 224x224

ℹ️ INFO: This model is ready to use on Synaptics Astra Machina boards. An NPU optimized version of the MobileNet V2 1.0 224 Quant is installed in the Astra SDK Image.

Deployment on Synaptics Astra SL1600 Series

This particular model is compiled for Synaptics Astra SL1680 processor. You can find this model already pre-installed on Machina™ Dev kit with SL1680 processor.

You can also find the same model compiled for Synaptics Astra SL1640 processor pre-installed on Machina™ Dev kit with SL1640 processor.

Synaptics Astra Machina™ is Modular developer kit for Astra SL-Series of high-performance IoT processors with integrated Synaptics Veros™ wireless connectivity solution. Learn more here

Application binary

The synap_cli_ic command line application available in the Synaptics Astra Machine SDK makes it easy to run image classification models like MobileNet V2 1.0 224 Quant.

Inputs:

• The converted synap model (.synap extension)
• One or more images (jpeg or png format)

Outputs:

• The top five most probable classes for each input image provided

Command line usage on Astra SL1680 and SL1640:

MODELS=/usr/share/synap/models/

cd $MODELS/image_classification/imagenet/model/mobilenet_v2_1.0_224/mobilenet_v2_1.0_224_quant

synap_cli_ic -m model.synap ../../sample/goldfish_224x224.jpg

Example output on SL1680:

Loading network: model.synap
Input image: ../../sample/goldfish_224x224.jpg
Classification time: 1.79 ms
Class  Confidence  Description
    1       18.99  goldfish, Carassius auratus
  112        9.30  conch
  927        8.70  trifle
   29        8.21  axolotl, mud puppy, Ambystoma mexicanum
  122        7.71  American lobster, Northern lobster, Maine lobster, Homarus americanus

ℹ️ INFO: Input images are automatically resized to the size of the network input tensor by the software pre-processor. This is not included in the classification time displayed.

Performance on NPU

Processors	Inference Time (ms)
SL1680	1.79
SL1640	2.31

Optimize and Customize the model

Advanced users may wish to customize the source model and recompile it for the Synaptics Astra NPU. Please refer to the Bring Your Own Model section for more information.

For reference, the .synap format model provided on the firmware image was compiled for the Synaptics Astra NPU with the following .yaml settings:

outputs:
  - dequantize: false
    format: confidence_array class_index_base=-1

License

The source model is licensed under Apache License 2.0.

The compiled model for on-device deployment is covered under the Synaptics Astra EULA.

Learn More

• Synaptics AI Developer Zone: Get started with documentation, tutorials and resources for your Edge AI journey.
• Astra Support Portal: Connect with our engineering team and community.