xBerry Case studies Digital Touch-Up

Digital Touch-Up

automated professional photography for car dealers

Watch the video

The modern era has shown us that a single image can effectively convey a message more potent than a lengthy explanation. How do the qualitative visuals translate into better sales? And how can we automate this process for individuals who may not possess professional photography skills? Together with our client*, we have come up with a solution to this challenge.

Scroll pageg icon

Challenge

Car shopping? Whether new or pre-owned, we all love to get close up and personal. We peruse official website photos, compare colors and models, and then take a closer look in the showroom or scrutinize photos before making a decision.

 

With these considerations in mind, our partner approached us with a unique business idea. Their objective was to enhance the efficiency, simplicity, and effectiveness of the automobile sales process for users of their car sales platform by introducing some innovative automation technology. 

 

As a forward-thinking company, xBerry is dedicated to discovering and incorporating innovative solutions and advanced technologies to drive our clients’ success. In line with this commitment, we started to develop some new functionality for the application, designed to automate the selection of filters to optimize the visual quality of car photos uploaded by users to their sales listings.

Goals

Our goal was to enhance our client’s app with a feature that would enable its users to optimize the visual quality of car photos added to their sales listings via automatic filter selection.

The initial concept of having a dedicated team manually processing these photos was deemed inadequate for this age of automation and efficiency.

 

To address this, xBerry created a solution that would automatically analyze each photo added by users, presenting them with three optimized options based on color, lighting, and image quality. The user could then choose the best option according to their preferences and add the final photo to their sales listing.

 

Moreover, at xBerry, we trained a machine-learning model to identify the most favored aesthetic preferences based on users’ selections, thereby continuously improving our filter recommendations to better align with the tastes of the app users.

Solution

  • For the backend, we used Python as our primary programming language due to its compatibility with machine-learning algorithms.

  • For our production environment we chose AWS in the form of IaaS mainly for its availability as well as simplicity of implementation and maintenance.

  • For the front-end, we opted for React to develop an intuitive and user-friendly interface.

  • To create our machine-learning model, we employed TensorFlow to streamline the development process.

Results

The successful integration of automatic photo retouching has proven to be a cost-saving and time-efficient solution for our client. The use of machine learning continually improves the app’s ability to select the optimal photo editing, resulting in improved user satisfaction and increased sales.

The photo editing process has been streamlined and now takes no longer than 30 minutes. To this day, the app has processed over 2 million photos (and still counting!).

 

*The details and name of this project are confidential and protected by a non-disclosure agreement.

Tech Stack

Python
Flask
PyTorch
Docker
Jupiter
React
AWS
TensorFlow

Planning a digital project?

Contact us Arrow icon