Automatic Billing System Using Quarky and Object Detection
by theSTEMpedia in Workshop > Science
76 Views, 1 Favorites, 0 Comments
Automatic Billing System Using Quarky and Object Detection
The Automatic Billing System is an intelligent, automated billing solution developed using PictoBlox and its Machine Learning Environment. The system is designed to simplify the billing process in small shops, supermarkets, or school canteens by automatically identifying products and generating bills without manual entry.
In this project, a camera captures the image of a product, and the Machine Learning model trained in PictoBlox identifies the item based on its visual features. Once the item is recognized, the system retrieves the product name and price from a predefined database and automatically adds it to the bill. The total amount is then calculated and displayed on the screen in real time.
The system demonstrates how Artificial Intelligence and computer vision can be applied in retail environments to make billing faster, more accurate, and efficient. It also reduces human errors during manual billing and improves the overall customer experience.
Supplies
Tools required to create this project are:
- Pictoblox Software: Download PictoBlox | Windows, MacOS, Linux, Chromebook, Android & iOS
- Laptop/PC
- Inbuilt or External Camera Setup
- Objects to showcase on camera - Quarky, Evive, Watch, Adaptor etc.
Open the PictoBlox Software & Choose Blocks Coding Environment.
Blocks coding environment is also known as scratch coding.
Add All Three-extension Mentioned Below.
Click on Open "ML Environment" From the Machine Learning Block Palette.
Click on Create New Project
Write Your Project Name, Select Project Type As Image Classifier, & Click on Create Project
Rename the Class Name According to Items & Add Image Samples Using Webcam
To Add More Items, Click on Add Class and Once All the Items Are Added, Click on Train Model to Train Your Customized Model
Once the Model Is Trained, Export Your Model in the Blocks Coding Environment
Click on Make a Paragraph & Add All the Necessary Details Mentioned Below in the Image
Customize the Style of Your Paragraph According to the Need, Preview It, & Click on Ok.
Make All Paragraphs As Shown in Picture
Click on Make a Variable From the Variable Block Palette & Enter the Variable Name
Create five variables like Count, Cus_Mobile, Cus_Name, Price, and Total.
Click on Make a List From the Variable Block Palette & Enter the List Name
Create three lists: Final price, Items, Quantity.
Add These Code Blocks in the Scripting Area of Tobi
Click on Choose a Sprite & Upload Two Customized Sprites As "next" & "finish"
Note: You can download the sprite image from here
Select the "next" Sprite and Add the Following Code Blocks in the Scripting Area
Select the "Finish" Sprite and Add the Following Code Blocks in the Scripting Area
Select the "Scan" Sprite and Add the Following Code Blocks in the Scripting Area
Expected Outcomes
- The system will automatically recognize products using the trained Machine Learning model.
- It will generate the bill automatically without manual product entry.
- The system will display product name, price, and total amount in real time.
- It will reduce human errors that normally occur in manual billing.
- The project will demonstrate the practical use of Artificial Intelligence and computer vision in retail systems.
- It will help users understand how ML models can be trained and deployed using PictoBlox.
Challenges & Solutions
Challenge 1: Incorrect Product Detection
Problem:
The ML model may sometimes misidentify products if images are unclear.
Solution:
- Train the model with multiple images from different angles.
- Ensure proper lighting conditions when capturing images.
Challenge 2: Similar Product Appearance
Problem:
Products with similar packaging may confuse the ML model.
Solution:
- Add more training samples for each product.
- Include distinct labels or markers on products.
Challenge 3: Camera Image Quality
Problem:
Low-quality camera images reduce detection accuracy.
Solution:
- Use a good-quality camera or webcam.
- Maintain proper distance and focus while scanning items.
Challenge 4: Limited Dataset
Problem:
The ML model performs poorly with very few training images.
Solution:
- Increase the number of training images for each product.
- Train the model with various lighting conditions and orientations.
Future Scope
- Integration with barcode scanning for hybrid billing systems.
- Addition of an IoT-based cloud database for storing billing data.
- Development of mobile app integration for digital receipts.
- Expansion to large supermarket automation systems.
- Integration with robotic checkout systems using devices like Quarky STEM Robot.
- Implementation of a voice-based billing assistant using AI.
Conclusion
The Automatic Billing System demonstrates how Artificial Intelligence and Machine Learning can simplify traditional billing processes. By using image recognition through PictoBlox, the system automatically identifies products and generates bills efficiently.
This project highlights the potential of AI-driven automation in retail environments, reducing manual effort and improving billing speed. It also helps learners understand the practical implementation of Machine Learning, computer vision, and automation technologies in real-world applications.
Overall, the project showcases how smart billing systems can transform the future of retail and customer service.
The sb3 code file for this project can be downloaded from here