
Learn: The Philosophy of Artificial Intelligence
Learn moreNow pizza makers can rejoice: the process of making the most delicious pizza has become much easier. Programmers, this is our merit. Modern technologies allow you to create your favorite Italian dish using only fresh ingredients and controlling the baking process. Thanks to this, the dough is thin, the crust is crispy, and the toppings are juicy and flavorful. Just the way we like it.
AI = QC
Large franchise pizza chains have long used neural networks to control the quality of their products. This approach allows for the effective monitoring of preparation standards and ingredients, ensuring high-quality pizza and customer satisfaction. Integrating modern technologies into production and delivery processes helps improve service and increase the brand's competitiveness in the market.
In 2019, Domino's Pizza Australia implemented an innovative system called DOM Pizza Checker. This system uses artificial intelligence based on neural networks and machine learning to evaluate the appearance of a pizza. With this technology, it is possible to quickly determine whether a product meets the company's high quality standards. The implementation of the DOM Pizza Checker demonstrates Domino's Pizza's commitment to improving the quality of its products and enhancing customer service.
Dragontail Systems developed a unique program designed to optimize restaurant operations. The development process took three years. The main component of the Pizza Checker is the intelligent QT AI Camera, installed above the tables where finished dishes are cut and packaged. This innovative technology significantly improves the quality of customer service and speeds up kitchen processes, allowing restaurants to effectively manage their business and improve the user experience.
The image processing algorithm analyzes pizzas based on thousands of photographs. It recognizes the type of pizza, studies the composition of the toppings, and evaluates the uniformity of the topping and cheese distribution. If the finished dish does not meet established quality standards, the DOM Pizza Checker system will notify you and prevent low-quality food from being sent to the customer. This guarantees high quality service and customer satisfaction.
With the onset of the COVID-19 pandemic, the program was adapted to comply with epidemiological measures. Algorithms now monitor the presence of gloves and masks among employees, as well as the cleanliness and regularity of the cutting table. If sanitary standards are not met, the neural network immediately notifies the restaurant manager of the violation. This ensures a high level of safety and protects the health of both workers and customers, which is especially important during the current pandemic.

Customers can be confident that their pizza is prepared in compliance with all food safety and hygiene standards. The DOM Pizza Checker system provides continuous monitoring of the cutting table and regularly reminds about the need for cleaning and disinfection. This ensures the high quality and safety of our products.
Don Meij, Managing Director of Domino's Group in Australia, shares his vision for the company's development and its role in the market. In the face of increasing competition and changing consumer preferences, Domino's strives to adapt to the challenges of the times. Meij emphasizes the importance of innovation and technological development to enhance customer service. He emphasizes that the company is actively working to improve its delivery processes and expand the range of dishes offered. Domino's also places special emphasis on sustainability, implementing environmentally friendly practices in its operations. This not only strengthens its market position but also attracts the attention of conscious consumers concerned about the future of the planet.
The implementation of Pizza Checker has been considered highly successful. In the first year of use, over 50 million pizzas were scanned, resulting in a 15% increase in quality scores. These statistics confirm the effectiveness of the Pizza Checker in raising product quality standards in the pizza industry.
The system has been successfully implemented in 850 restaurants across Australia and New Zealand. Domino’s also intends to launch the Pizza Checker in North America, which will improve service quality and increase customer satisfaction.
Installing the system in a restaurant costs between $500 and $1,000. You also need to factor in a $50 monthly subscription fee, which provides access to the company’s computer vision neural networks on the server. This investment can significantly improve the efficiency of the establishment and improve the customer experience.
The pizza chain Papa John’s is testing a new technology to assess the quality of its products. The company’s division, operating in the CIS and Central Europe, has developed an innovative system called Pizza Magnifico (PizzaM). This system automatically analyzes the quality of each pizza coming out of the oven, which helps ensure a high standard of service and customer satisfaction.
To train the artificial intelligence, the developers used a dataset consisting of more than 700 thousand photographs of Papa John's pizza, collected over a period of 10 years. Before loading into the neural network, each image was rated by the company's experts on a scale of 1 to 10. This approach allowed us to create a high-quality model capable of recognizing and analyzing the visual characteristics of pizza, which significantly improves the user experience and increases the quality of the services offered.

PizzaM's artificial intelligence provides control over the pizza preparation process, ensuring that every customer of the restaurant chain Papa John's in Russia, the CIS, and Central Europe will receive a high-quality product. Thanks to innovative technologies, PizzaM optimizes all stages of production, ensuring the high standard of signature pizza in every location.
Our team is proud to have collaborated with leading Russian experts in artificial intelligence and computer vision in the project's development. We have implemented their cutting-edge technologies to enhance the quality of our pizza, ensuring excellent taste and customer satisfaction. Innovations in AI and computer vision help us optimize processes, improve recipes, and create unique offerings for our guests.
Christopher Wynn, President of Papa John's Western, emphasized in an interview with Horeca-magazine the importance of innovation and service quality for success in the foodservice industry. He emphasized that continuous process improvement and attention to consumer preferences are key factors driving the company's growth. Wynn also emphasized the need to adapt to changing market conditions and utilize new technologies to improve the customer experience. The Russian chain Dodo Pizza is actively using artificial intelligence to monitor the quality of its products. Instead of developing a separate device, the Dbrain team of programmers utilizes the built-in cameras of modern smartphones, which are readily available to everyone. This streamlines the monitoring process and ensures a high standard of quality, which positively impacts customer satisfaction and company efficiency.
The program was developed for the brand's community of mystery shoppers, which numbers over 50,000 members. These shoppers conduct covert monitoring of the quality of dishes in the franchise's restaurants, ensuring high standards of service and customer satisfaction.
The system functions as a Telegram bot that allows users to rate pizza. To receive a rating, reviewers take a photo of the purchased pizza and send it to the neural network. In response, the bot will provide a rating on a ten-point scale. This approach simplifies the pizza quality verification process and makes it more accessible to users.
The artificial intelligence was trained in several stages. In the first stage, the neural network was provided with 50,000 pre-labeled pizza photographs taken from various angles. In the second stage, the developers engaged mystery shoppers to further train the neural network, allowing them to evaluate the results of the artificial intelligence. This process ensured greater accuracy and efficiency in image recognition, which is important for the further application of the technology in various fields, such as the restaurant business and food delivery.
Based on numerous user feedback, the neural network adapted its internal parameters to more accurately perceive the quality assessment system used by humans. This allowed it to effectively replicate similar mechanisms in its work. The process of training and testing artificial intelligence took a month and a half.

Cooperation with Dbrain made it possible to develop a tool that effectively determines the quality of tests based on mystery shopper reports. This tool will form the basis for maintaining high pizza quality while scaling internationally, which in turn will help reduce quality control costs.
Fedor Ovchinnikov is the founder of Dodo Pizza, a well-known pizzeria chain that is actively developing its franchise. He has become a symbol of successful entrepreneurship in the food service industry. Ovchinnikov applies innovative approaches to business, which allows the company to remain competitive in the market. Dodo Pizza strives to provide customers with high-quality products and services, which contributes to its growth and recognition. Fedor Ovchinnikov inspires young entrepreneurs by sharing his experience and knowledge of running a business and managing franchises.
Always fresh ingredients
Domino's Pizza in Malaysia and Singapore intends to implement blockchain technology to ensure quality control of its products. The company has already signed a collaboration agreement with SingularityNET, founded by David Hanson, creator of the famous robot Sophia. The use of blockchain will enable tracking of every stage of the product lifecycle—from harvesting to processing and packaging—which will increase transparency and consumer trust. This innovation will be an important step in ensuring the high quality of food and the safety of Domino's Pizza products.
In preparation for the implementation of new solutions, Dodo Pizza is actively using artificial intelligence to optimize ingredient supply management in its restaurants. A specially developed program predicts which products and in what quantities each pizzeria will need over the next week, which helps reduce costs and minimize leftovers. This solution contributes to increased restaurant efficiency and improves customer service.
The implementation of the automated system has significantly reduced the time employees spend on manual calculations, from five hours per week to a minimum. Previously, employees often made errors, leading to shortages or excess inventory, which in turn caused spoilage and additional costs. Automated processes have improved calculation accuracy and optimized inventory management.
Modern artificial intelligence systems analyze a multitude of factors that are impossible for humans to comprehend. They can identify and account for hidden sales trends, such as seasonal demand fluctuations and the impact of holidays. This allows for the optimization of business processes and more informed data-driven decisions. Using such technologies, companies can increase their competitiveness and better meet customer needs.
The new development has improved procurement forecasting accuracy by 18% and significantly reduced employee workload. This allows each franchise restaurant to save an average of 100,000 rubles monthly. In 2021, the artificial intelligence system was implemented in 50 cafes and restaurants. As a result, neural networks helped reduce costs by 54 million rubles over the course of the year. The implementation of such technologies not only optimizes processes but also increases the profitability of the food service industry.
The forecasting system used at Dodo Pizza was developed by the Russian branch of the Norwegian company Crayon, a long-standing Microsoft partner in the field of artificial intelligence and machine learning. This solution is based on the powerful Microsoft Azure Machine Learning platform, ensuring highly accurate and effective forecasts. This approach allows Dodo Pizza to optimize business processes and improve customer service.
The task presented an interesting challenge. After a thorough analysis of the data, it became clear that there are a significant number of different classes of ingredients, each with unique consumption patterns. A separate forecasting model had to be developed for each class, which required a deep understanding of their characteristics and features. This approach ensured more accurate results and allowed us to better take into account consumer preferences.
When we began working on the project, we were promptly provided with all the necessary data, the Microsoft Azure cloud infrastructure, and integration solutions. Thanks to this, we were able to quickly implement the project, completing all work in a week and a half.
Vladimir Eronin, Director of the Artificial Intelligence and Data Analysis Competence Center at Crayon Russia, notes the importance of integrating modern technologies into business processes. He emphasizes that the use of AI-based solutions contributes to increased efficiency and cost optimization. In the face of rapid technological progress, businesses must adapt to new challenges to remain competitive. Integrating AI technologies into the company's strategy opens up new opportunities for data analysis and improved customer service.
Dodo Pizza plans to soon implement the new system across all Dodo Brands divisions, both in Russia and internationally. A key step will be full integration with the information systems of ingredient suppliers, which will streamline processes and improve the company's efficiency. This innovation will strengthen Dodo Pizza's position in the market and ensure a higher quality of customer service.
A Neural Network as a Chef
Artificial intelligence has made significant progress and is now capable of creating new recipes. As part of the project "How to Generate (Almost) Anything in the World," MIT students trained a neural network to generate original pizza toppings. This project demonstrates how modern technologies can influence gastronomy, opening up new horizons for culinary creativity and inspiration.
The project used the textgenrnn recurrent neural network, available on GitHub. This open platform, created by data scientist Max Wolf, is designed to analyze and generate text information. Using textgenrnn allows for the efficient processing of text data and the creation of new texts, making it a valuable tool for researchers and developers in the fields of machine learning and artificial intelligence.
The system was fed numerous recipe texts from cookbooks and online resources. Using this data, textgenrnn was able to create unique recipes, sometimes so original that even an experienced chef would have difficulty imagining them. These generated recipes can inspire new culinary experiments and open new horizons in gastronomy.

Artificial intelligence has an excellent ability to integrate various sources of information, which leads to unexpected results. In our experiment, we observed a similar phenomenon: the AI selected a combination of shrimp, Italian sausage, and jam, which was mentioned in a dessert pizza recipe. This highlights the creative potential of AI in creating unique culinary combinations, which can inspire new gastronomic experiments.
Pinar Yanardag leads the "How to Generate (Almost) Anything" project, which was featured in Inc. Russia. The project focuses on innovative approaches to generating ideas and solutions in various fields. Under Yanardag's leadership, the team is working on creating tools and methods that optimize the process of generating content and ideas, making the project relevant to professionals in marketing, business, and technology.
Artificial intelligence not only suggests interesting combinations of pizza ingredients but also sometimes creates names for non-existent products. For example, a neural network can suggest a unique "Ranch walnut dressing," highlighting the creativity and originality of the AI's approach to dish creation. These unusual ideas can inspire you to experiment in cooking and add variety to your usual menu.
Chef Tony Nasser of Boston's Crush Pizza creatively reimagined the results of artificial intelligence and selected five of the most promising pizza toppings. These innovative combinations not only bring freshness to the menu but also demonstrate how modern technology can inspire culinary art. As a result, pizza becomes more than just a dish, but a true gastronomic experience that impresses guests with a variety of flavors and textures. Tony Nasser is confident that these toppings will attract the attention of both regular customers and new visitors looking for a unique gastronomic experience.
- blueberries, spinach, feta cheese;
- bacon, avocado, peach;
- shrimp, Italian sausage, jam;
- sweet potato, beans, brie cheese;
- apricot, pear, cranberry, ricotta.
I made each of these pizzas, making adjustments to the recipes as I went, and invited the project participants to try the new dishes. The most delicious and original was the pizza with shrimp, jam, and Italian sausage. Nasser decided to add it to his restaurant's menu, which confirms the popularity and uniqueness of this dish.
Dodo Pizza has launched an interesting experiment called Dodo AI-pizza. Experts in neural networks from MIPT and Skoltech participated in the development of the project. This approach allows the company to use advanced technologies to improve the quality of service and optimize the pizza preparation process. Dodo AI-pizza demonstrates how artificial intelligence can transform the traditional catering industry, offering customers new opportunities and increasing business efficiency.
This domestic development stands out in that it takes into account the molecular compatibility of the original products. To achieve this goal, two neural networks were used. The first network was trained on 300 thousand recipes from all over the world, which allows it to take into account the diversity of culinary traditions. The second network analyzes the results of molecular compatibility studies conducted at leading universities in the US and UK. This approach provides more accurate recommendations for ingredient combinations, making the cooking process not only more efficient but also improving the quality of the dishes.
Dataset collection proved to be the most challenging aspect of working with machine learning. Currently, there are no ready-made databases, which forced us to automatically process large culinary websites. However, given that recipes on these sites are published by ordinary users, various issues arise, such as typos and inaccuracies in the text. These errors can significantly complicate further data processing and require additional refinement to ensure the quality and accuracy of the information.
Egor Baryshnikov is the developer of Dodo AI-pizza, known for his innovative platform for automating pizza delivery processes. According to TechInsider, his work significantly improved the company's efficiency by implementing advanced artificial intelligence technologies. Baryshnikov continues to contribute to the development of automation, making Dodo AI-pizza a leader in the food delivery market.
To optimize recipes using neural networks, programmers used the Word2vec algorithm. This algorithm converted pizza ingredients into a vector representation, allowing the model to effectively identify foods that pair well with each other. This approach significantly improves the quality of recommendations and allows you to create unique combinations of ingredients for making pizza.

Pizza created according to the neural network recipe received The name "Open Source." Its filling featured ten ingredients in unusual combinations, including perch fillet and melon. This culinary masterpiece was available for tasting at the Dodo restaurant in Moscow on October 10, 2019. The Dodo AI-pizza team has posted the source code, a description of the methodology, and several generated recipes on the GitHub platform, allowing other chefs and culinary enthusiasts to experiment with innovative pizza-making ideas.
AI Draws Pizza
In 2019, researchers from the US and Qatar presented a paper on a platform capable of generating realistic images of pizza based on a text recipe. This technology allows consumers to visually assess the appearance of the dish before it is prepared. During the development of the platform, the team led by Dima Papadopoulos used generative adversarial networks, known as GANs. The use of such technologies opens up new horizons in cooking, allowing not only to inspire cooking, but also to improve the user experience in choosing dishes.

Pay attention to additional materials:
How artificial intelligence creates images of cats: discussing generative adversarial neural networks (GAN)
Artificial intelligence has become an important tool in the field of image generation, especially when it comes to creating cute and funny cats. This technology is based on generative adversarial networks, or GANs. These networks operate on the principle of competition between two models: a generator and a discriminator. The generator creates new images, and the discriminator evaluates their authenticity, determining whether they are real or generated.
The GAN training process involves many iterations, during which the generator improves its skills, and the discriminator becomes increasingly accurate in its assessments. As a result of this competition, the neural network can generate images that are difficult to distinguish from real cat photos.
The applications of GANs are not limited to the creation of images of pets. This technology is finding its way into a variety of fields, including art, fashion, and even medicine. However, it is the creation of realistic images of cats that attracts a wide audience and demonstrates the potential of AI in creativity.
Thus, generative adversarial networks represent a powerful tool for creating unique content that opens new horizons in the world of art and design.
A neural model created in 2014 has proven itself as an effective tool in the field of image generation. With its help, artificial intelligence has learned to create photorealistic images of animals, including cats, and generate fake avatars for non-existent people. Furthermore, the model is capable of imitating the style of paintings by great artists, opening new horizons in art and design. This technology continues to evolve, offering ever-higher quality and realism in the creation of visual content.
Papadopoulos presented his development, called PizzaGAN. This neural network has the unique ability to step-by-step add ingredients to an image of a pizza and accurately simulate the cooking process. PizzaGAN opens new horizons in image generation, enabling the creation of visualizations that capture not only the appearance but also the texture of the finished dish.
Additionally, an image of an uncooked pizza can be uploaded to the program, and it will generate an image demonstrating how the dish will look after baking. The result includes realistic cheese drips, a crispy crust, and crispy bacon bits. This allows users to visualize the final result and enhance the appeal of their dish.
With the click of a button, PizzaGAN can add layers of olives, pepperoni, or fresh herbs to an image of a finished pizza. The result will look as if it were a real photograph, which makes the technology especially impressive.

The program can distinguish between 12 different pizza toppings: arugula, bacon, broccoli, corn, basil, mushrooms, olives, onions, pepperoni, peppers, pineapple, and tomatoes. You can experiment with your choice of ingredients on the project's website. To ensure the ingredients are distributed naturally across the pizza's surface, the neural network was developed using layer mask generation technology similar to that used in Photoshop. By removing the ingredient layer, PizzaGAN can accurately recreate the appearance of the hidden toppings, providing a realistic representation of the finished dish.
To train the artificial intelligence, Dima Papadopoulos' team used two different image sets. The first set consisted of pizza clip art, and the second included half a million real photos collected from social media using the #pizza hashtag. This approach ensured diversity in the model's training, allowing it to better recognize and interpret pizza-related visual elements.
The scientists used a supporting neural network to process the photographs, removing irrelevant images. The final dataset retained only those pizza photographs that were taken from the correct angle and were of high quality. This improved the accuracy and efficiency of visual data analysis.
The final dataset contains approximately 9,000 images, each accompanied by a text annotation. This 2.8 GB archive is available for download.

The PizzaGAN model has demonstrated good results, but it is still not perfect - the accuracy of determining the correct order of ingredients is 88%. Researchers from Rutgers University in New Jersey, USA, continued the work of their predecessors and significantly improved the model's performance. Thanks to new methods and algorithms, it was possible to increase the efficiency of ingredient sequence recognition, which opens new horizons for the application of artificial intelligence in cooking and other fields.
The researchers analyzed the Papadopoulos dataset, identified errors in it, and made the necessary corrections. The updated and improved dataset is called Pizza10.
Secondly, an improved version of the StyleGAN2 generative adversarial network from NVIDIA was used. This tool is called MPG (Multi-ingredient Pizza Image Generator). MPG creates high-quality pizza images by combining various ingredients, significantly expanding visualization capabilities and aiding in the development of unique recipes. This technology opens new horizons in the fields of culinary arts and digital marketing.
We developed a web application that allows users to specify pizza creation parameters and feed them into a neural network. As a result, users receive a unique pizza image generated by the StyleGAN2 model. This application also provides the ability to evaluate the improvement in the image quality generated by StyleGAN2 over previous models.
The MPG neural network was trained on four NVIDIA K80 GPUs, each with 12 GB of video memory. This process took six days, during which the algorithm generated 2.4 million images of fictitious pizzas. The use of powerful graphics processors allowed for faster training and improved image quality, making this project an interesting example of the use of neural networks in content generation.
The scientists presented their research results in a scientific paper, and the source code for the MPG project is available on GitHub. A demonstration of the application was also made in the form of a video.
Developments that allow you to see the pizza before it is cooked have begun to be actively used for commercial purposes. Domino's Australia has once again stepped up its game with the introduction of a smartphone app called New Pizza Chef. The app includes a visualization feature that allows customers to see what their future pizza will look like. This approach not only improves the user experience but also helps boost sales, as customers can visually preview their options before ordering.
An application using augmented reality (AR) technology gives users the unique opportunity to create their own pizza, choosing from a wide range of toppings and crusts. To do this, simply select the desired products from the proposed list. This innovative development makes the process of choosing ingredients simple and fun, allowing everyone to create the perfect pizza to their taste.
A virtual image of a pizza, created from the selected ingredients, is instantly displayed on the screen of a mobile device. The program uses the built-in smartphone camera to analyze the surrounding space, allowing the user to visualize a model of the dish right on their dining table. This provides a unique opportunity to “try on” a pizza before ordering, which makes the selection process more interactive and enjoyable.

The application allows you to visualize the pizza of your dreams, providing the opportunity to examine it from different angles.
Michael Gillespie, Director of Digital and Technology Affairs at Domino’s Group, shares his views on the development of digital technologies in the food delivery industry. In his interview published on cmo.com.au, Gillespie emphasizes the importance of innovative solutions for enhancing customer experience and streamlining business processes. He notes that the implementation of new technologies not only improves the quality of service but also contributes to sales growth and strengthens the company's position in the market. In the highly competitive food delivery industry, attention to digital strategies is becoming a key success factor for companies like Domino's.
To increase the visibility of your content in search engines, it is important to optimize the text with keywords and phrases. Make sure your text answers user questions and satisfies their needs. Use keywords naturally to improve readability and avoid spam.
If you want to attract more readers, focus on creating unique and valuable content that answers relevant queries. Analyze your competitors and determine which topics are of most interest to your target audience. Update your information regularly to stay relevant and keep users interested.
By following these recommendations, you can improve your content's SEO and make it more accessible to your audience.
- Inhuman cuisine: how AI learned to cook and became a chef at a Moscow restaurant
- Test: AI on the table - oversalted on the back. Recognize what dish the computer invented
- Website parsing with Python and the Beautiful Soup library: a simple three-step guide
The Philosophy of Artificial Intelligence
You will receive answers to important questions about "machine thinking" and understand the role of philosophy in the development of AI. Identify the key differences between natural and machine intelligence, conduct relevant scientific research, and present in public.
Learn more
