Visually coherent banners generated in multiple sizes and designs by just setting few parameters

Automated Banner Generation System

The Client

Food Delivery Startup

With a network of over 50,000 restaurants, one of the most prominent food delivery startups that deliver food items across hundreds of cities throughout India wanted an automated banner generation system for their offers.

Akaike’s edge-cloud agnostic solutions offer great flexibility.

Flexibility 0%
Cost 0%

Executive Summary

Industry Overview


How the world eats is changing dramatically. A little under two decades ago, restaurant-quality meal delivery was still largely limited to foods such as pizza and Chinese. Food delivery has become a global market worth more than $150 billion, having tripled since 2017. In the United States, the market has more than doubled during the COVID-19 pandemic, following healthy historical growth of 8 percent. Succeeding in the fast-growing food-delivery ecosystem will require understanding how overlapping economic forces affect a complex web of stakeholders.
Business Challenge

Customer Experience

Visuals attract the audience’s attention far more than just plain text. The food delivery startup was looking for an automated banner generation system to make the Call-to-Action (CTA) more productive as the existing banner generation required a significant amount of manual effort from an army of graphic designers, which was not only expensive but also time-consuming and prone to errors. The customer wanted banners to be generated automatically by just setting parameters like CTA button, food item, discount, etc., without any specialized software that did not need some technical expertise to operate. As per these parameters, the image would be displayed to the target set of audiences and visually coherent banners would be generated in multiple sizes, designs, and styles with strict business guidelines.

The Akaike Edge

Inbuilt libraries, DL models with transfer learning capabilities

Experienced ML and DL Ops teams

Efficient Deployment 0%
Integration 0%
Ongoing Maintenance 0%


Inbuilt libraries, DL models with transfer learning capabilities
Step 1.

Requirement Gathering

For the task, the primary step was to gather all the guidelines for the banner generation such as image quality, background color, CTA text color and font, placement of the food item, and text on the banner.

Step 2.

Conducted computer vision experiments for different steps

Since there was an exhaustive set of guidelines and rules for the banner generation, we divided the steps into smaller tasks. For example, to find the appropriate food image as per the food item, we experimented with image stock APIs and selected the most relevant one. Additionally, to get the precise location of the food items, object detection API was used to get the right bounding box for the food item to adhere to the guidelines. Similarly, the algorithm was designed to put the text on the left and the food image on the right side of the banner as per the guidelines.

Step 3.

Created a Web Application

A web application was created wherein the users can input parameters such as food, discount, CTA, and more and they get 3-4 banner samples from which the most relevant one was used.

Research shows that 87% of Vision AI
projects do not yield expected results

either owing to training data insufficiencies stalling the project, or being too slow to deploy. Our AI experts can help you accelerate in data sparse environments.

We use cookies to give you the best experience.

Get in Touch