Woman lost 312 Pounds in One Year After Dying

It all began when Deborah and her husband Donny decided to transform their life, and only twelve months later, a miracle occurred. They’ve changed into super inspiring weight loss role models for…

Smartphone

独家优惠奖金 100% 高达 1 BTC + 180 免费旋转




Vison Based AI Electronic Scale

Introduction:

Smart electronic weighing scales are widely utilized in supermarkets and neighborhood fruit/vegetable shops to enhance work efficiency. These devices integrate weighing and POS (point-of-sales) software, streamlining operations in shops.

However, current smart electronic weighing scales face certain challenges:

1. The system requires manual item selection on the scale, which can be time-consuming for shoppers unfamiliar with the operation.

2. Incorrect item selection leads to pricing errors.

Fig1. A normal Smart Electronic Scale

To address these issues and further enhance the efficiency of the weighing process, a vision-based AI Electronic Scale solution has been introduced to the market.

Additional hardware needed: AI camera.

Operational Process Flow:

1. Place the items on the scale.

2. Once the scale detects the weight, the camera activates and captures an image of the items on the scale.

3. The system utilizes item identification algorithms to identify each item.

4. The software calculates the total price by multiplying the unit price with the weight. It then either prints a label, generates a receipt, or proceeds with the payment process.

Item Identification Model Creation:

1. Load the image data of items with labeling.

2. Convert the image to grayscale.

3. Apply morphological opening to enhance features.

4. Utilize a CNN (Convolutional Neural Network) approach to extract features.

During Operation:

1. Capture the image when the weighing result stabilizes.

2. Convert the image to grayscale.

3. Perform item identification.

Fig3. Vison Based AI Electronic Scale using Android OS (Device picture from iMin technology)

Discussion:

In practical scenarios, the device may not have internet connectivity, necessitating an offline identification algorithm.

Moreover, the shop may sell items do not present in the existing dataset. In such cases, a transfer learning model can be employed, allowing shoppers to load actual item images for feature extraction using a pre-trained model.

Add a comment

Related posts:

Convert Leads into customers.

To convert your leads into your customers you need to follow some steps if you successfully captured a lead.. “Convert Leads into customers.” is published by Lucky Gupta in Lucky Gupta.

How Evermos is building the largest social commerce company in Indonesia

All great businesses begin with a problem that won’t go away. A problem often so deeply ingrained in a system that no one wins but everyone just accepts it. So it was with the retail industry in…

What Can be Tokenized? The Future of Security Tokens

It may be hard to believe, but soon you could earn a fraction of Cristiano Ronaldo’s earnings. You could own a few square metres of an apartment in Hong Kong or perhaps a stake in a collectible…