I am asked more and more by my clients and prospects if we can help them with identifying, analyzing, and tagging image and video files in their enterprise. This is not surprising given that the modern enterprise is creating and storing an ever-increasing amount of videos and images either as standalone files or contained inside documents, PDFs, and PowerPoint presentations.
Stored images and videos files, by nature, are usually very hard to discover by employees. A standard full text scan gives back no information and there is very rarely any description/metadata about the files. Compounding this problem, images and videos can be found deep in the pages of a document or presentation. Depending on the company, those files are often very important assets that help users do their jobs. If someone can’t find a “how to” video to fix an issue or the latest company logo, then this can be a big problem.
Fortunately, there are AI solutions available that can empower users to much more easily find the image and video files they need. The function of an AI classification offering should:
Identify what or who is in the video/image file. Does the video/image contain a cell tower, oil rig, car design, signature, or the CEO’s keynote speech? This identification becomes part of the metadata of the file, making it easier to find.
Extract speech within a video and create a text searchable transcript with meaningful metadata.
Capture all the text within an image. This text could be model numbers, captions, road names, etc. which helps improve search relevancy and quality.
Determine what activity is taking place in an image/frame. Repairing an engine part, writing on a whiteboard, raising a hand, etc.
Recognize a person or expert from a repository of facial images.
Improve accuracy of recognizing objects, activities, and scenes over time.
To be clear, I am not advocating that search be focused on video or images. Rather, I am advocating that new technologies make it possible for a single search to be able to find the information a user needs, independent of what the format is. This should be the case whether it is a Word document or a PDF, a conversation in Yammer, Jive, Confluence or Teams, an image by itself, or an image in a document, or a video.
Let’s be honest about it, employees today have a growing expectation for internal search to be “Google”-like. A search on Google will result not only in webpages, but also in images and videos. In the corporate environment, if a user needs to find how to prevent their company-owned laptop from crashing, then the answer may only be found in a “how to” video hosted in ServiceNow. Traditional, document-only search would not be helpful in this example, leading to end user frustration and loss in productivity.
Organizations need to start thinking that there should be no boundaries for information when it comes to internal search. Relevant information is independent of where it is stored and what type of content it is. With an AI-driven classification system in place, documents, images, and videos will be on equal footing and you can empower your employees to find critical information quickly. Without an automated solution, the growing number of videos and images stored in the enterprise will be another great unknown trove of valuable knowledge that is “locked up” with no access from the outside.