Intelligent Weighing Scale For Fruits and Veggies Mart

The development of artificial intelligence has brought great convenience to our day-to-day life. Over the past decade, many retailers have briskly integrated a spread of technological applications. One of the increasingly used technologies is that the delivery of service through self-service technol

Project Title

Intelligent Weighing Scale For Fruits and Veggies Mart

Project Area of Specialization

Artificial Intelligence

Project Summary

The development of artificial intelligence has brought great convenience to our day-to-day life. Over the past decade, many retailers have briskly integrated a spread of technological applications. One of the increasingly used technologies is that the delivery of service through self-service technologies. Self-service intrigue to the customer’s desire for more pliability, independence and freedom when shopping and enables them to command their own product choices and paces. Meanwhile, self-service opens up opportunities for grocery retailers, to relieve the pressure on their often-over-stretched workforce, reduce labour and training costs, and optimize workflows and operational processes.

Different techniques have been used over the past years for fruit and vegetable recognition using Computer vision technology. Because of the wide diversity of food and vegetable types, image recognition of food and vegetables is very difficult. Our project presents a novel approach to food detection and recognition is using deep learning algorithms. We aim to make an intelligent scale for fruits and veggies marts based on the idea that since weighing is more routinized for sales staff they can serve more customers faster and significantly reducing the risk of over traffic at the self-service scales during busy shopping times. Fruits and vegetables will automatically be identified through built-in camera.  The touch screen will allow users to instinctually navigate the system without worker assistance or memorize PLU numbers. Retailers who implement self-service concepts ease their employees' workload, reduce costs on labour and training, and benefit from higher productivity and greater customer satisfaction.

Project Objectives

Complex and time consuming self-service and manual systems can make customers choose another grocery store. Since customers are the reason companies survive, their satisfaction is the businesses’ key to success. The need for a system to reduce processing time exists because of their constant endeavour to save time.

The objective of this project is to improve the identification process of fruits and vegetables performed by the self-service system in the markets and to improve the usability by using graphical user interface compared to existing manual systems. More specifically, the improvement should consist of a faster process and more user friendly system. The purpose of implementing computer vision is to simplify the object identification process by moving the process from human to a computer. Also the user will get rid of remembering all the prices and weighing procedure they are doing manually these days. This will minimize the error by removing human factor and will have all the records stored on your machine.

Project Implementation Method

Detection part: (Hardware: Raspberry pi 4, External USB camera or Pi-Camera)

• As our project is based on object detection and recognition, first of all we need to select a best object detection model, we are using yolov4-tiny, a real time object detection model which is the light version of yolov4 and can easily run-on single board computers like Raspberry pi, Jetson Nano, VisionFive RISC-V etc.
• We need the data set for training this model, we have chosen 10 classes for now i.e., potato, tomato, onion, chilli, turnip, cucumber, carrot, garlic, apple and guava. We are making dataset by ourselves and collecting images of these items form different marts and grocery store. Then we are annotating these images to make our dataset ready for the training our model.
• As our target amount (at least 500 images per class) of dataset is achieved. Then we will feed this data to the model for training.
• For training model, we will need GPU, so we will use google Collaboratory which is providing us free GPU, we will train our model over there.
• Then we will connect the USB camera using USB port or Pi-Camera using CSI port of the raspberry pi for detection of the item placed on the scale.
• Finally, we will deploy our model on the raspberry pi and check the results.

Weighing part: (Hardware: hx711 module, load cell 5kg)

• We will make an arrangement just like normal scales used in the market, we will use a load cell inside this scale for measuring the weight of the items placed on the scale. The item placed on the scale will be detected (Detection part) and along with their weight and will be shown on the graphical user interface. We are using an hx711 amplifier along with the load cell. The hx711 amplifier will be connected to raspberry pi through the GPIO pins. It will sense the weight and will show us on the Screen.

User Interface part: (Hardware: LCD touch screen, small portable printer)

• In this part we will make a graphical user interface (GUI) through which the user will be easily interact with the scale.
• We can easily make an interactive graphical user interface in python using tkinter library.
• It will make the user able to change the price of the items, user can see the bill detail, can save and print the bill for customers.

Benefits of the Project

The intelligent weighing technology offers many benefits to food businesses and they are as follows:

Reduces the possibility of error:

Our intelligent weighing machines will help perform all the weighing operations precisely without any human error. There is no need to rely on operator to record data in the system. The data recorded by smart weighing machines are precise and free from human error. This will directly minimize the financial loss.

Speed up the process:

The manual system is slow. The automated system is very fast and quick to execute commands. This will reduce the pressure on employees and minimize waiting time at checkout.  In this way self-service weighing will become easier for retailers too.

Improvement in store image and higher customer satisfaction:

Our intelligent weighing machine will add elegance to the sales counter and retail stores. Moreover, with the innovative design store image will improve and consequently the customer will be highly satisfied.

Eliminates fraud:

The automated systems ensure that all the processes go fair and eliminates the chances of fraud and human error. By investing in small weighing systems you can be sure to do fair business.

Technical Details of Final Deliverable

Final Delivery of the project is the hardware/software integrated system.

Software:  

• In software we will have a python script for running yolo-v4 tiny object detection model. This model is simple and can easily run-on single board computers like raspberry pi, nano jetson, VisionFive RISC-V etc.
• Another python Script that will do interaction with hx711 module and will measure the weight of the items placed on the scale.
• And one python script of graphical user interface (GUI) through with user will be easily interact with the scale using touch screen for different purposes like changing prices, printing bills, checking records etc.
• All these python script will be combined in a single python script and can also runs in parallel.

Hardware:

• Parts: Raspberry pi 4 kit, hx711 amplifier, load cell 20 kg, 7-Inch Capacitive Touch LCD Screen, small portable printer.

• In hardware we choose raspberry pi 4 with 8gb ram and faster quad-core CPU, which can easily run yolov4- tiny object detection model which is most computationally expensive part in the sense of computation. Raspberry pi 4 gives us round about 2.5 fps speed which is enough in our case.
• Raspberry pi has GPIO pins through which we can easily interact with the environment using sensors and actuators. Using GPIO pins we will interact with our hx711 amplifier and load cell for measuring the weight. This part will be just like traditional scales in the market.
• The user will interact with the scale through the display. The Raspberry Pi Official 7-Inch Capacitive Touch LCD Screen is available for raspberry pi. The 7” Touchscreen Monitor for Raspberry Pi gives users the ability to create all-in-one, integrated projects such as tablets, infotainment systems and embedded projects.
• For printing bills for customers, we will use a portable printer. We can use POSX  891 Thermal Receipt printer 80mm having USB and Serial Interface.
• For giving power to the scale, we are using power adopters with come along with a raspberry pi kit and touch lcd screen kit.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Food

Other Industries

Education , IT

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Decent Work and Economic Growth, Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com