Akida examples

To learn how to use Akida, the QuantizeML and CNN2SNN toolkits and check the Akida accelerator performance against some commonly used datasets please refer to the sections below.

For more self-contained projects covering training, conversion and deployment on Akida hardware, check out the BrainChip DevHub repository.

General examples

Start with either of the two workflow guides below, depending on the framework you train in: the Global Akida workflow for TF-Keras models, or the PyTorch to Akida workflow for PyTorch models (through ONNX).

Global Akida workflow

Global Akida workflow

PyTorch to Akida workflow

PyTorch to Akida workflow

AkidaNet/ImageNet inference

AkidaNet/ImageNet inference

DS-CNN/KWS inference

DS-CNN/KWS inference

Age estimation (regression) example

Age estimation (regression) example

Transfer learning with AkidaNet for PlantVillage

Transfer learning with AkidaNet for PlantVillage

YOLO/PASCAL-VOC detection tutorial

YOLO/PASCAL-VOC detection tutorial

Segmentation tutorial

Segmentation tutorial

Quantization

Advanced QuantizeML tutorial

Advanced QuantizeML tutorial

Upgrading to Akida 2.0

Upgrading to Akida 2.0

Off-the-shelf models quantization

Off-the-shelf models quantization

Spatiotemporal examples

Gesture recognition with spatiotemporal models

Gesture recognition with spatiotemporal models

Efficient online eye tracking with a lightweight spatiotemporal network and event cameras

Efficient online eye tracking with a lightweight spatiotemporal network and event cameras

Edge examples (Akida 1.0 only)

Akida vision edge learning

Akida vision edge learning

Akida edge learning for keyword spotting

Akida edge learning for keyword spotting

Tips to set Akida edge learning parameters

Tips to set Akida edge learning parameters