Aidlux Industrial Vision Defect Inspection

Aidlux Industrial Vision Defect Inspection

1. Introduction to AidLux

AidLux is a fusion architecture platform independently developed by Chengdu Ajiaxi Intelligent Technology Co., Ltd. It provides Android/Hongmeng + Linux fusion system. The dual systems can be used independently and communicate with each other.
Ajiaxi is committed to the continuous innovation of artificial intelligence core technology, creating an original integrated architecture operating system to meet the needs of various application scenarios, providing an out-of-the-box Al tool chain to make it easier to use chip computing power, and realizing the rapid migration and implementation of AloT projects. Comprehensively assist enterprises in industrial intelligence, AI education, robotics, smart home and other fields to achieve rapid product iteration and intelligent upgrade.

2. yolov8 model conversion based on AidLux

1. model conversion

• Open the website: http:/aimo.aidlux.com/
• Enter the trial account and password:
  Account: AIMOTC001, Password: AIMOTC001
• Select the corresponding original model

2. Model conversion TFLite

• After the model is optimized, the before and after models can be compared.
• Download the optimized model

3. Deployment of yolov8 model based on tflite

1. Prepare

• images folder: inference image samples
• onnx original model: yolov8_slimneck_SIOU.onnx
• tflite: py file of tflite model and inference

2. Model deployment

• Open AidLux on your mobile device
• Log in to AidLux on the PC web page. The default user is root and the password is aidlux.
• Put the images, tflite model files and tflite.py files into the home directory
• Open vscode, open the tflite.py file, right-click and run in the terminal. You can find the inference results in the result.
• You only need to install python-opencvi and numpy, and there is no need to install torch, making inference as simple as possible.
• Try to choose a lower version of python-opencv, such as version 3.4, otherwise opencv will report an error.

reason

• Try to choose a lower version of python-opencv, such as version 3.4, otherwise opencv will report an error.

Finally, here is the address of the video on the effect of industrial visual defect detection based on the Aidlux platform.

https://www.bilibili.com/video/BV1uH4y1D7Vf/