Skip to main content
Radarly: Image Recognition
Cheyenne V. avatar
Written by Cheyenne V.
Updated over a week ago

Thanks to our proprietary image recognition technology 'Computer Vision'. You can measure, understand and optimize the impact of your brand.

This functionality allows you to detect the following on an image:

  • Logo

  • Scenes

  • Number of people and their ages

  • People gender

  • Objects and animals

  • Celebrities

This article will cover:


Accessing this Feature in Radarly

From the main view of the Posts & Analytics page, click on the tab at the top centre of your screen to switch views and access the Image Recognition (May also be called Computer Vision)


Logo Recognition

Logo recognition on an image allows you to understand what is being said about your brand but also to capture photos where your brand appears without being mentioned.
By including it in your analysis, logo recognition will allow a fine measurement of the performance of a campaign or product launch and bring new points of view of analysis.

Performance measurement of the sponsorship investment:

With logo recognition, all advertising campaign media, on social networks can be detected, measured, quantified and analysed.


The technology allows for the prediction of the logo presence, even if the logo is fuzzy, partial or in the background, we can know the real ROI of investment.

Search for insights through the understanding of consumption environment:

If a young girl posts a photo of her new mascara applied to her face, Radarly will be able to detect it.

Note

  1. This feature is a paid add-on. Get in touch with your Account Manager to find out more.

  2. In case the Logo or Celebrity is already vetted by our software, back data is available until the 15th of March 2024.

  3. Data capture can take place based on a Logo or Celebrity in an image.


Top Scenes in Images

On your project, the images are analysed to be described in the most relevant way possible.

Thanks to the 'keyword cloud', you can understand what the images in your project are about.

You can display the Top scenes in images:

  • By setting a personalized view in the workspace settings (need to be an administrator)

  • By modifying a view of the “Posts & Analytics” page to display the “Top scenes in the images” in the word cloud (click on the 3 dots at the top right of the graph, then click on “Keywords” and choose “Top scenes in images from the drop-down list)


Age Detection

It is now possible to filter the publications according to the ages detected on the images.
The age qualifications of the people are as follows :

  • Mostly children

  • Mostly adolescents

  • Mostly young adults

  • Mostly adults

  • Mostly seniors

You can view this filter in two ways:

  • By adding a new widget in one of the views of your workspace through the project settings (need to be an administrator)

  • By changing the data displayed in a widget ( though the 3 small dots at the top right of each widget)


Celebrities Recognition

Our Artificial Intelligence can now detect an image of a celebrity.


Activating Image Recognition on a Project

This feature is a paid add-on. Get in touch with your Account Manager to find out more.

To activate celebrity recognition, you need to follow three steps:

  1. You need to collect 300 photos of the face of the person you want to recognize.

  2. Put all these images in a folder and download them on a drive.

  3. Write an email to help@meltwater.com to make the demand, and don't forget to add the link to the images folder.

Note:
These images must follow the following rules:

  • Each of the 300 images must be unique.

  • There should only be one face in each picture.

  • The 300 photos must show the person's front and side views.

** There are now 325 celebrities on which you can capture posts without having to give images as examples


💡 Tip

Need more help? Feel free to reach out to us via Live Chat or check out our Customer Community.

Find answers and get help from Meltwater Support and Community Experts.


Did this answer your question?