STILINGUE Eyes May 06, 2024 12:18 Updated Index: Vision beyond reach How it works? How to view? Image transcription Collected by AI Visualization How far can you see? Expand the scope of your research with the functionality for text reading in images. If you want to further expand the possibilities of data collection and visualization in your research, STILINGUE Eyes can help. This feature was developed to enhance vision through text extraction from images posted on social networks and other digital media using Computer Vision—a form of Artificial Intelligence (AI) that provides new ways to process and analyze information. Vision beyond reach The concept of Computer Vision arose from the need to bring to computers the human-like ability to understand and interpret images and videos. Currently, the applications of this field are diverse and contribute to advancements in areas such as Medicine, Engineering, and facial recognition technologies. How it works? STILINGUE Eyes is the exclusive Computer Vision engine created to identify, extract, and classify text from images in open sea publications—meaning those brought in by the Listening area. This feature acts as a complement to the AI engines used to analyze texts in publications. When using STILINGUE Eyes, the texts extracted from these images will be "read" and organized according to the searches used in the Groups created in the Search Configuration. For example, if the system identifies the term "airport" in an image, and this term is included in a Group in your search, it will be classified within that Group. Additionally, the functionality can collect a publication exclusively by the text in the image, meaning that even if the publication's text does not contain any words that would cause the publication to be collected, it will be collected if it is present in the image. This means that STILINGUE Eyes offers broader coverage to your research. Another example: the text "Christian Dior" was identified in the image of a publication, which also allowed the collection of that publication. The text of the publication did not match the searches in its Groups, but the text extracted from the image refers to a topic of interest to your research that would not have been possible to investigate without reading the image. See below an example of a publication collected based on the text extracted from the image: How to view? To access publications collected by STILINGUE Eyes, you must first open the Publications Filter in Listening. When opening this area, you will have various options to refine the publications displayed on the Wall, such as selecting only publications with images (an interesting feature to compare the total number of images present in the search with those that had texts extracted by the functionality). Read the guide for more information. The viewing options are located below the sentiment icons within the "What" area and are activated with a click on the checkmark icon. They are: Image transcription When activating this option, the filter will return all posts with images containing text extracted via Artificial Intelligence. Here, it's important to note that posts indicated by this filter may not necessarily have been collected based on the content of the image; the collection process may have occurred through the text present in the post, for example. Regardless of how it was collected, the publications that had text identified in the images by AI will be shown. Collected by AI In this option, the filter will only show publications captured by Artificial Intelligence through the extraction of text in the image. The main difference here compared to Image Transcription is that the publications returned in this filter were effectively collected and added to the search due to the fact that the text extracted from the image matched the searches of the Groups. If the extracted term is among the searches of the Groups that make up your search, it will be collected according to these Groups. Additionally, an extracted term that matches both Groups and Themes will be categorized within the Theme. This way, it is possible to obtain the collection of publications that would not be in the search without the action of STILINGUE Eyes. One way to explain these differences would be to think of each universe of posts as a set within another, where the content is segmented. Within the larger set are all the posts from your search. In a smaller set, we have posts with images, and in turn, publications related to STILINGUE Eyes that had some text extracted from the images: those filtered by Image Transcription and finally, Collected by AI. We can say that Image Transcription is a more comprehensive filter than Collected by AI since the former includes not only publications whose collection was made possible by the text extracted from the image. You can use both options separately or together, according to the needs of your search. But if both are selected, remember that the less comprehensive filter (Collected by AI) will prevail. Visualization After selecting the desired option, the publications will be displayed on the Publications Wall just like other types of content collected through Listening. Regardless of the option chosen (Image Transcription or AI Collected), when clicking on the publication card, you will see the text extracted by Artificial Intelligence displayed below the original text, under the heading "Text extracted from the image (via AI)": Another possibility is to view the terms extracted by Zoom (which can be accessed within the Publications Filter) or in other areas of Listening, such as Correlated Terms and other charts that are part of the Discoveries menu: How far can you see? STILINGUE Eyes is a powerful feature to enhance your analyses, but there are some points to consider regarding its use: Sentiment: the text extracted from the image will not determine the sentiment classification of a publication. This means that sentiment remains related only to the original text of the publication. Textual search: the text extracted from the image is not considered in textual searches, only the original text of the publication. Carousel: in the case of carousel-type content (with multiple images in the same publication), STILINGUE Eyes can extract and display texts from all images present in the carousel. To navigate through the carousel images, use the arrows located on the sides of the image, on the publication card, or simply drag the image to the side. Learn more about the functionality in the FAQ. For more information, visit the discussion on the subject at our community or videos on our channel. 😃 Related articles FAQs STILINGUE - Eyes Blip Desk Overview Sonora STILINGUE Creating interactive messages in WhatsApp Registration of Tags by Line