Home Blog AI in Digital Asset Management

AI in Digital Asset Management

The use of AI in digital asset handling and management seems to be evolving rapidly. This article explains how we can already use AI for dealing with repetitive tasks, and some of the opportunities we see on the horizon.

First, we start with - What is a DAM?

Digital Asset Management (DAM) tools are software solutions that handle and store digital assets. These assets might be images or video. They might be .pdf catalogues, guides, and resources. They could be files used by designers in Photoshop, Illustrate, or InDesign.

Benefits of AI in DAM

Some of the benefits already apparent for using AI in DAM include being able to:

Remove or reduce the risk of human error
Remove or reduce the time spent on manual data entry and tagging
Make it easier for contributors to upload assets to the collection without needing to be trained in the DAM tool
Analyze massive collections of images and files

Opportunities for AI in DAM

Artificial intelligence in DAM is very much in its infancy, yet there are constant improvements being made to how AI can improve DAM. The big questions are:

Does AI improve DAM?
How much can AI do to improve DAM?
Can AI make handling large collections of assets easier?
Is AI smart enough to deal with variations and anomalies in images?

At this stage, most DAM solutions employ the use of a third-party AI tool to do the work, rather than it being a core capability of the DAM solution itself. However, some tools on the market do have AI features inbuilt.

AI can be used to assess the quality of, or improve or even write product descriptions. This capacity could have benefit when you have images of many hundreds of products that require a description before they can be listed online.

The biggest opportunity for AI in DAM is surely through the use of metadata and tags to catalogue, compare, and update references in images. AI does indeed offer an opportunity for reducing the time spent on manually adding metadata.

Tagging of images using AI

The tagging of images enables you to sort collections and find what you are looking for quickly and easy. Many DAM systems will begin adding tags as soon as assets are imported, based on what is depicted in the image. AI can be used to expand tagging capabilities and match more accurate tags to each asset.

Apple Photos already uses AI to tag photos in your Iphone. Type a term like ‘beach’ or breakfast into the search bar and you’ll see any images in your collection with images of your seaside holiday or your smashed avo on toast. In these cases the photos program is using its own data to assess the image, and can’t alter or add any key words of your own.

How else could AI be used in a DAM tool?

The following are some further ways AI could be used within a DAM to automate processes and improve asset management. AI could be extremely useful for e-commerce providers who have large collections of assets that need to be assessed and edited to meet style and brand guidelines.

Blur detection and quality control

AI can assess overall quality of an image, the resolution and if there are issues with blurriness, scanning large collections for any images that might not meet your organisation’s quality or clarity standards.

Object identification and removal

AI could be used to assist with identifying sensitive or offensive images or aspects within images. This could be extended to include the blurring out of components of an image. AI can help moderate collections of assets, in line with pre-applied standards or community guidelines.

Background removal and editing

Machine learning algorithms can enable AI to identify and then separate out a main subject from any background images. It does this by identifying and labeling each pixel as being in either part of the foreground or the background. This could also be used to remove all backgrounds from images, to they display products against a white or colored background only.

Facial, celebrity and landmark recognition

Facial recognition identifies when the same person or people appear in multiple images, once it has been tagged and identified in an initial image. Celebrity recognition on the other hand, does not rely on an existing tag of a person to be able to make identification. Landmark recognition occurs when AI recognizes and automatically tags a site of importance.

Colour recognition and matching

AI could help carry out color matching and identification on images which could be helpful when honing in on assets to be used in a catalogue. Another example of how color matching could be useful would be searching your collection for examples of a particular color, though its unique color ID code. In this way, branded assets could be easily collated, or you could have ready access to any sub-branded materials or, even look for deviations in file setup.

What are the weaknesses of AI in DAM?

AI is not a replacement for your DAM. There is a risk of overreliance on AI in the DAM environment- it’s best to consider how it can enhance or improve capability rather than seeking to take away the need for the DAM in the first place. AI can be used to automate basic information input, but it is limited in terms of assessing the relevance and accuracy of metadata, or usefulness of the assets in the first place.

How can AI improve video editing and distribution?

AI can be used in video to assess video quality and relevance. This includes assessing long stretches of footage to scan for particular parts or scenes which meet a particular description. AI can also perform video edits by adding color balance, handle scene transitions and even match music to activity on screen.

How can AI be used for subtitling and translations in video?

AI-powered speech recognition algorithms can convert the spoken words in a video into text. This transcription can then be used for generating closed captions or for easier content searching. AI can also manage translations and generate subtitles in multiple languages.

What are the weaknesses of AI in video?

While some of these functionalities sound truly impressive, it’s important to remember that the results may vary greatly, and may not leave you with a product you are proud of. AI is in its element when it is performing repetitive, consistent tasks that don’t involve outliers, nuance or an understanding of context. As such, it might be most useful in helping identify where there are assets with quality issues, helping you to assess large collections and decide where the work needs to be done.

The other thing to consider is the highly emotive nature of video, and how it is such an important part of your brand management. Your staff are much more likely to have a good sense of how to construct a meaningful, impactful piece of video about your business that AI.

The tool you use for AI in images and video also make a big difference, so make sure you assess the options and go with the best you can afford.

Examples of AI in asset handling

Amazon Rekognition is a product of Amazon Web Services that can add image and video analysis to your applications. It can identify the objects, people, text, scenes, and activities, or any inappropriate content from an image or video. Users at G2.com rate the product highly for ease of use, but flagged some problematic attributes, with one user saying “Real time large scale implementation is bit of a challenge with SaaS. At times, the software goofs up by not making the right accurate prediction articulation.” Cost was also mentioned in reviews, and this product can get pricy when used to scale.

Digital Asset Management in Pimcore

Pimcore’s DAM tool enables you to:

Escape data siloes- by integrating your DAM within the existing IT business environment
Achieve brand consistency- by removing inaccuracies and identifying unauthorized variations to brand and style guidelines
Deliver seamless customer experiences with high-quality and uniform assets across all channels

Pimcore DAM has the capacity to handle and manage large collections of assets, across just about any file format. It has a wide range of contemporary features such as:

VR/360 images
Automatic AI/ML face and hotspot recognition
Image editing
Versioning, tagging, scheduling, and more