In times of home office and digital meetings, we often lack a face to our conversation partners.. Video conferences are prohibited in many companies due to GDPR, work council resolutions or due to technical restrictions. Some employees upload self created profile pictures to the communication platforms. Most companies also do not have standardized profile pictures in outgoing email signatures, on the intranet and on business platforms such as LinkedIn.
What it does
In a perfect world, the profile photos in all channels correspond to the company's corporate design rules and the employee can update them any time with minimal effort. With our developed solution the employee sends a selfie to a defined email address without worrying about the background or the image size. Our solution edits this input image in several GDPR compliant steps.. We make good use of artificial intelligence to remove the background.. We resize and crop the image based on facial landmarks.. We brand the image to the company's specifications and optimize it for each target platform.. We send back the set of profile images to the employe..
You want to test it?
Send a selfie from your upper body to this email-adress: email@example.com Confirm the GDPR double opt in email. Shortly after that you will receive your branded profile images by mail.
How I built it
I always try to combine existing solutions to create new business models. In this case we use AI facedetection and AI background removal from cloud service providers. With the information from those services, we are able to shape any selfie from a user into a set of corporate design conform profile images for Intranet, eMail signatures, communication and colaboration tools and so on.
Challenges I ran into
One issue was the inconsistency of smartphone cameras when it comes to rotation information. To make sure, that the head is always up, we build an image proxy that replaces the exif information and rotates the image to the right orientation.
Accomplishments that I'm proud of
I am extremly proud of the team. Eduard an Mladen worked hand in hand with me to reach our goals. 3 People from 3 different countries worked together like we are sitting next to each other in the office.
What I learned
Even something small can have a huge impact. Companies spent big money to keep their company design consistent everywhere. With this little tool, each employee can be part of this effort.
What's next for Employee Image Automation
In order to advance the project, we need contacts to companies that would like to incorporate this feature.
The roadmap includes additonal output options like Biometric passport photos.. Also the creation of Company ID Cards for employees and visitors will be possible.. An API for easy integration into an existing IT landscape will be provided as well..
Lets talk about money
The variable costs per use essentially consist of the use of the cloud resources. Here, fee-based services from GDPR-compliant third-party providers are used for artificial intelligence to remove the image backgrounds and for face recognition with a suitable crop of the image. Since these services run as a pay per use model, it is not possible to offer a flat fee to clients.
The billing of the service is therefore processed via a credit system. Interested companies can buy prepaid credits and use them if necessary. Purchased credits are valid for 12 months. Credits not used during this period expire if no new credit bundle of the same value is purchased.
The price per credit will be like this: 2,50 € 0 - 1.000 Credits 2,25 € 1.001 - 5.000 Credits 2,00 € 5.001 - 20.000 Credits 1,50 € 20.001 - 100.000 Credits 1,00 € > 100.001 Credits
Scalability of the service
Since we are using cloud services, there are no limits in scalability of the service The current infrastructure is able to process about 500 images per minute => 720.000 per day If one component is low on resources, a queue mechanism will take care about out.
Please feel free to contact me at firstname.lastname@example.org