Summary of Contents for brainchip Akida x86
- Page 1 Akida x86 Evaluation Kit - User Guide Akida x86 Evaluation Kit User Guide P a g e...
-
Page 2: Table Of Contents
Akida x86 Evaluation Kit - User Guide Table of contents INTRODUCTION ..............................3 SYSTEM INFORMATION ............................3 SYSTEM SETUP FOR RUNNING DEMOS & THE AKIDA BOARD EVALUATION ..........3 INITIAL SYSTEM CHECK AND SETUP ........................4 AKIDA DEMOS ..............................5 ........................... -
Page 3: Introduction
Akida x86 Evaluation Kit - User Guide NTRODUCTION This PC-based system is setup for Akida NSoC ADK1000 testing and evaluation. The Akida evaluation board is already installed and tested on the Shuttle PC. The Miniconda virtual environment is installed along with several demo applications. All demos can run on or the Akida hardware seamlessly. -
Page 4: Initial System Check And Setup
(https://doc.brainchipinc.com/examples/) but already installed on your machine (see paragraph 7) Code your own python program in our conda environment or creating another one, including your favorite libraries And don’t forget you can get support from Brainchip using Akida customer support portal: https://support.brainchip.com/portal/en/signin P a g e... -
Page 5: Akida Demos
Akida x86 Evaluation Kit - User Guide KIDA EMOS 5.1 R UN THE KIDA EMOS 1. Open a terminal window, navigate to /home/bcdemo/akida_demo 2. Enter: conda activate akida The system is now ready to run the demos. 5.2 A KIDA DEMO This application uses a MobileNet-like model for "Keyword Spotting"... -
Page 6: Akida Classification Demo
Akida x86 Evaluation Kit - User Guide 5.3 A KIDA LASSIFICATION DEMO This application uses the MobileNet/ImageNet model from akida-models to classify Webcam frames. This demo requires a USB camera. Enter python classification_demo.py Place objects in front of the camera for prediction. -
Page 7: Edge Face Recognition Demo
Akida x86 Evaluation Kit - User Guide Figure 5: Edge learning demonstration window EDGE FACE RECOGNITION DEMO This application uses MobileNet models replacing the last layer to learn new classes from the Webcam frames. This demo requires a USB camera. -
Page 8: Detection Demo
Akida x86 Evaluation Kit - User Guide DETECTION DEMO This application uses a YOLO model from akida-models to detect objects or faces into webcam frames. This demo requires a USB camera. Enter python detection_demo.py Face, person, or object detection should occur with bounding boxes as well as scores. Two modes available: Car/person and Face. -
Page 9: Application Settings
Akida x86 Evaluation Kit - User Guide 6 APPLICATION SETTINGS Press F1, a menu will display > Application Settings. Figure 8: General settings window 6.1 S WITCH FROM ARDWARE TO SOFTWARE BACKEND Switch HW / SW: Select “HW device” for Akida HW backend, then > Close. -
Page 10: Turn On Display About Performance
Akida x86 Evaluation Kit - User Guide Figure 10: Application window background changes depending on chosen backend The screen background color is blue when Akida demo is running with the software backend while it’s grey when running with the hardware backend. -
Page 11: The Akida Metatf Ml Tools
Akida x86 Evaluation Kit - User Guide TF ML T KIDA OOLS MetaTF is installed with the examples in the following directory. Open a terminal window navigate to /home/bcdemo/MetaTF Activate environment entering: conda activate akida_env Run examples with Jupyter notebook or python. -
Page 12: Shuttle Pc Re-Installation
• Follow the instructions you will find in the README.md file. 3 – Install MetaTF Please refer to Akida online documentation (https://doc.brainchipinc.com/installation.html) to install MetaTF. 4 – Akida demos To include Akida Demonstrations in this reinstall, please contact Brainchip to get the latest release. P a g e | 12... -
Page 13: Troubleshoot
Akida x86 Evaluation Kit - User Guide ROUBLESHOOT 9.1 R UNNING ISSUE If the system is stuck or if the demos do not run after performing the required steps in this user guide, try soft reboot sudo reboot If the above doesn’t work do a full hard power down reset the system and start over.