https://store-images.s-microsoft.com/image/apps.46036.d20a9bd6-5a14-4761-92dc-b3b1d0329095.09b9d88b-d15d-434d-b028-d385fb18ff03.c9ca4bb4-1fc8-4d67-b7b5-a432877ee402

AIoT Enabled AOI Service

от MegAAOI

A POC to demonstrate how AI ML can do to remedy your quality inspection hassles at your factory

For the corporate and enterprise in the manufacturing industry, due to the high demand of customer's quality requirement, defect inspection becomes a crucial last mile in achieving total customer satisfaction, as well as light-off smart manufacturing of your factory.

Prior to engaging full-scale implementation to improve your inspection performance, the service can provide you with a POC against your AOI application. You can assess how well the AIoT enabled AOI solution can contribute to your application to facilitate your next move, which is your implementation decision. At the end of this 6 weeks service, you will get a validated AI model trained by the image data (photo) acquired from your existing AVI/AOI machine.

The offer comes with initial AI model training with follow-on 3 times of AI retrain close-loops based on the human-verified Underkill and Overkill data provided by your skilled inspection personnel. The project activity includes:
  1. Project inbrief: 0.5 week as the kickoff meeting
  2. Initial AI model training: 1 week for data acquisition (NG: 500/200 and OK: 500/200 photos for both training and lab testing respectively), cleaning, training, and lab test
  3. Initial AI model validation: 1 week for customer's skilled inspection personnel to provide Underkill and Overkill data after validating those OK and NG inferred by the initial AI model. The test data shall be provided from the live production line (at least 1000 from the inspection machine that previous as the source of training/lab testing data)  
  4. AI model retrain: 1 weeks each, total 3 weeks for 3 retrain close-loops. The retrain close-loop will be triggered once the manual-validated Underkill and Overkill data reaching 100/100 respectively. The test data shall be provided from the live production line (1000 photos per upload, may need a few uploads to trigger retain). An IoT mechanism will be provided to the customer to forward the manual-validated data to the offerer for retraining. The offerer will adopt Microsoft Azure cloud services to facilitate the AI retrain close-loops
  5. Project outbrief: 0.5 week including outbrief data preparation 

Бърз преглед

https://store-images.s-microsoft.com/image/apps.60382.d20a9bd6-5a14-4761-92dc-b3b1d0329095.09b9d88b-d15d-434d-b028-d385fb18ff03.dfc55ac3-fa3e-4115-95ee-a027d94e6b7d
/staticstorage/5610156/assets/videoOverlay_7299e00c2e43a32cf9fa.png
https://store-images.s-microsoft.com/image/apps.33019.d20a9bd6-5a14-4761-92dc-b3b1d0329095.09b9d88b-d15d-434d-b028-d385fb18ff03.f28d6fe1-4f11-44a9-969e-397b38b50f1a
https://store-images.s-microsoft.com/image/apps.5962.d20a9bd6-5a14-4761-92dc-b3b1d0329095.09b9d88b-d15d-434d-b028-d385fb18ff03.0a1bda8d-9cd6-421a-a368-fb07b4daea55
https://store-images.s-microsoft.com/image/apps.47022.d20a9bd6-5a14-4761-92dc-b3b1d0329095.09b9d88b-d15d-434d-b028-d385fb18ff03.83df1197-545f-4eb6-abe7-8f113439c4d2
https://store-images.s-microsoft.com/image/apps.48683.d20a9bd6-5a14-4761-92dc-b3b1d0329095.09b9d88b-d15d-434d-b028-d385fb18ff03.db8d1f77-1ae5-4bdc-820b-277c938ab710