Table 21

Inspiration

源于在食堂就餐时，在自助取餐通道，拥堵点往往是位于整个流程的末尾段，人工收费的部分。当队伍越急越长时，负责收钱的阿姨甚至会紧张出错。因此，通过我们这套系统，可以保证快速和稳定，也可以减少相应的人力资源。

What it does

Canteen, a place needed to be every single day. It is hard to avoid mistakes during calculation by human. Using Computer Vision Tech can reducing human resourcing and improving accuracy. A simple demo to show a bigger opportunity.

How we built it

通过计算机视觉的方式，解决了餐厅内菜品价格计算的问题。通过一套简易廉价的装置，包含一个不需要太贵的芯片和一个不需要太清晰的摄像头、搭配一块透明板和一个补光灯，可以快速识别手写于餐盘底部的价格。

1. Take a picture.
2. Using Hough Transform to detect circle.
3. Base on the labeled circle, crop a square.
4. Rotating and Cropping and Whitening.
5. Recognize the number based on CNN. 6.Sum up and Output.

Challenges we ran into

• Recognize the circles in difference lighting environment.
• Build the predict model.
• Confirm the correct direction of the digit picture.

Accomplishments that we're proud of

• Use red line to mark direction.
• Use Mnist to train our model.
• Use Hough Transformation to detect circles in picture.

What we learned

• Basic machine learning & Computer vision.
• Tensorflow model used in python.

What's next for Table 21

Future Improvement

• A much cheap chip. (The price for this equipment will be less than 50 yuan.)
• A Much More Professional Bracket with supple lamps.
• Using number to present the variety of food.
• Using Big Data to detect the situation of nutrition.
• Future Application
• Corporate with detection of nutrition.
• Apply in unmanned supermarket.
• ……

Built With

• cnn
• python
• tensorflow