CountMyBoxes

Inspiration

We need to scan or take picture the inventory items one by one to get the barcode or the QR Code.

What it does

Using a single snap of camera, AI can detect multiple QR codes. With multiple QR codes detect, we can also count how many inventory items in that picture.

How I built it

Using ML model from https://aws.amazon.com/marketplace/pp/prodview-avylkh3xuqs3e?ref_=srh_res_product_title

Challenges I ran into

-At the moment only can recognize five out six boxes. Although I had increased the instance into ml.m5.12xlarge but the QR recognition ML couldn't detect all of those six boxes. -To recognize a single QR code need to take multiple pictures as Sagemaker return 500 error code. -Not all QR code can be read. -It took more than 30 minutes pending to get the Sagemaker Notebook instance to be ready. -Many failures batch transform jobs

Accomplishments that I'm proud of

What I learned

Need also allocate time and effort to do demo and presentation.

What's next for CountMyBoxes

Create mobile app and integrate it with camera

Built With

• jupyter
• python
• quantiphi
• s3
• sagemaker