Visual Content Moderation AI For Social Platforms and Marketplaces

Typical content moderation looks at text to monitor subject and themes that go against the policy of a site. Visual Content Moderation does the same for images and videos contained on a site. This has become of key importance with the growth of platforms that host user generated content.

Visual content is difficult to moderate and requires computer vision technology tuned for this specific task. Needs also differ as user content and social media platforms wish to screen for varied themes, such as nudity, weapons, violence, hate speech, and terrorism, while ecommerce and social commerce platforms look for things like trademark, copyright, and design infringements.

Content moderation has become crucial to protect the public and the platforms themselves. But the growth of visual content has brought with it its own challenges. These can be addressed with inclusion of computer vision in the moderation process. Content online is increasingly visual. Gaming platforms, messaging apps, social media apps, video sharing platforms and more all host a large and growing percentage of visual content. In some cases, visual content is at the very core of the platform. Many automated systems are only text-based, leaving humans to do most of the visual work. This can cause a number of issues including enabling more room for error and causing distress to employees or contractors. Visual-AI overcomes all these challenges, at levels of efficiency and cost that simply cannot be matched by humans.

Yes. Any popular image and video formats can be processed and checked, including streaming video formats.

Absolutely. One of the key advantages of VISUA’s Visual-AI technology is our extremely low latency real-time processing, which is ideal in live streaming applications or where you require content to be checked in real-time.

VISUA is a leading computer vision company and we have worked very hard to develop the most precise visual content moderation technology that can process content at massive scale. Our goal is to always provide the best visual moderation technology that can operate alongside text-based moderation systems rather than trying to be an all-round content moderation service.

There are four major reasons for incorporating computer vision into a currently human-based moderation process.

Speed: One averagely resourced Visual-AI system can do the work of hundreds and even thousands of humans. As the content volume and/or complexity scales you would have to hire ever more people to try to keep up with the task. Visual-AI can process images and videos in seconds and can handle millions of images per day, automatically blocking them or flagging them for human review, as appropriate. Additionally, Visual-AI never sleeps or takes breaks, allowing your moderation service to operate around the clock 365 days per year.

Trauma: There are numerous accounts by ex-moderators of having to view horrifically disturbing content. This causes immense distress and ultimately burnout. In some cases it has led to employees taking legal action against their employers. Visual-AI has no emotion and reviews content dispassionately, therefore avoiding these issues entirely.

Cost: Scaling a business that must rely on humans as a core part of the business process is ultimately flawed. Instead, Visual-AI can scale easily and quickly. Once trained on the specifics of what you wish to detect and block it can double its throughput in a matter of hours by adding additional processing and memory resources. Doing the same with humans would take weeks/months and cost exponentially more. Not to mention the issues around lay-offs when you wish to scale a specific operation back, whereas you simply turn-off a Visual-AI system if it is no longer needed.

Appropriateness: Overall, humans are not best equipped to handle the task of moderating content. Concentration and accuracy drop based on levels of fatigue, boredom, stress and burn-out. These factors do not affect a computer vision system, which is designed to handle massive volumes of visual content hour after day, day after day.

Absolutely not. It can be argued that humans are not the most appropriate resource to deal with massive volumes of visual content, but it can equally be argued that Visual-AI is not best suited when it comes to making decisions about edge case content. For instance, is an image ‘violent’ or is it journalistic content in the public interest? Does a video contain gratuitous nudity or is it an artistic piece. Where Visual-AI sees black and white, humans can see shades of grey, allowing them to make more informed decisions. Humans will therefore always be of key importance in making determinations in these edge cases. But instead of grinding through thousands of images, they are reviewing only a small number that require the higher-level intelligence that only humans can apply.

Social Media has arguably had the biggest impact on society in history since the investigation of the printing press. Yet it is also one of the most divisive innovations in history. The challenge comes from the ability of individuals and groups to post inappropriate, damaging and misinforming content at will. Combined with the massive volume and breadth of predominantly visual content content, social media platforms have been challenged to keep up.

With Visual-AI, hate symbols, offensive text (burnt into images and videos) and imagery, nudity, controlled objects and substances, can all be detected and immediately blocked or flagged for review by the content moderation team.

Ecommerce platforms are increasingly being held to task by legitimate manufacturers and rights holders. High-profile legal cases have penalised platforms for allowing counterfeit and copyright/trademark infringing products to be sold. Equally, sites that allowed products containing racist or hate speech to be created have seen large-scale and very public backlash.

Visual-AI can protect these platforms from these issues by checking products and designs uploaded to the platform before they go live. This dramatically reduces the strain on Trust and Safety teams, making it easier for platforms to stay compliant.

Live Streaming videos are particularly complex when it comes to moderation. You need to detect and block infringing content, but you can’t cause lag or delay in the stream. VISUA’s highly efficient computer vision technology added to its on-premise and on-device capabilities allow moderation of live content without degrading the user experience.

Users of video hosting services will often have agendas and needs that may go against the policies of the service providers. Content needs to be checked thoroughly to ensure offending content can be blocked before it goes live. But this requires advanced technologies that can check videos to a high frame-rate so that even fleeting infringements can be caught.

VISUA can check videos frame by frame in real-time to detect and block a wide range of infringing content.

Gaming platforms typically have numerous channels which require tight moderation, both inside and outside the platform. Many allow chats, image and video sharing, and custom skin creation, all of which requires moderation – especially where minors are often a large part of the user base. Visual-AI can detect and flag visual content that breaches the terms and conditions of the platform.

There are many reasons why some computer vision providers are better than others. It would be outside the scope of this answer to list them all, but the key reasons are as follows:

1) Most offerings are not purpose built for the task of visual content moderation. VISUA’s API was.

2) One important factor in moderating visual content is the ability to quickly and easily add new logos, marks and other visual themes to the library. Not all computer vision APIs allow this and even the ones that do require lengthy training. VISUA’s API is the only one with our patented Adaptive Learning Engine, which eliminates the need for training data in many cases or at least drastically reduces the volume of training data required. This allows you to react quickly to new content themes that infringe your content policies.

3) In applications looking for real-time processing you’ll need a computer vision solution that can be deployed on-premise. VISUA’s offering is the only one that supports this.

4) Batched task processing is essential for efficient API calls. It allows you to fire multiple tasks from a single API call rather than having to make multiple calls on the same piece of media. Only VISUA and Microsoft’s APIs provide Batched Task Processing.

We have also developed specific comparison documents, which are available in our Resources section. However, if you would like to discuss this further please don’t hesitate to get in touch.

Yes, you may well be able to build this yourself. But the key question is what are the pros and cons of doing so? You could build a CRM system for your business, but in virtually every case you’ll instead go use Hubspot, Salesforce or similar. Building your own CRM, although technically possible, provides no major advantage to your business.

Similarly, building your own computer vision solution, specifically to moderate visual content, also doesn’t make sense for the following reasons:

1. The problem exists today – you don’t have time to develop this tech, while infringing content is getting through undetected.
2. Visual-AI (computer vision) is a very specific discipline of machine learning and requires expertise that you likely don’t have.
3. Building a low scale solution is achievable – building a solution that can deal with the massive scale usually required in production environments is a much more complex proposition.
4. Building a Visual-AI solution that allows you to easily add new logos, marks and other graphical themes in minutes and doesn’t require massive volumes of model training data every time is not something you’ll have access to with off-the-shelf API solutions.
5. Currently available computer vision APIs do not provide on-premise capabilities – vital for low-latency implementation.

Great question. This is another quite unique offering from VISUA. Deployment can be implemented in the cloud, on-premise or even a combination of cloud and on-premise if required.

We like to think that our Visual-AI (computer vision) API is very easy to implement as part of any workflow, in fact, in most cases implementation takes as little as two hours. We have very clear API documentation also. But we are not simply an API provider, so do not hesitate to get in touch with any questions you may have. We also implement a very thorough onboarding process and as a client you will have direct access to our team for any ongoing support questions.

Absolutely! We have two options for you in this case:

Low-Code: Simply use our API to send us the media you need processed. Once processed we will add our findings into a dedicated dashboard for you to review. You can provide post/file/product details along with the files with the API and these will be added to the dashboard entries, allowing you to quickly click back to the relevant content on your site and take appropriate action.

No-Code: If you don’t have the ability to use our API then our No-Code offering allows you to provide a secure location for images and videos to be collected by our engine for processing. After processing is complete they are loaded into a dedicated dashboard for your review. Adding relevant post/file/product details as part of the content metadata allows us to populate this information alongside the dashboard entries, allowing you to quickly click back to the relevant content on your site and take appropriate action.

Yes, you can find very clear API documentation for our Logo Detection endpoint, or indeed any of our other technologies. You can find all relevant documentation here.

Absolutely! Unlike other solutions on the market that charge significant fees for support, or force you to reach out to third-party consultants, VISUA is proud to be much more than simply an API provider. You can get in touch with any questions you may have during your research and feasibility stage. We also implement a very thorough onboarding process and as a client you will have direct access to our team for any ongoing support questions.

VISUA is not a typical API company like other providers. As such we don’t provide ‘support packages’. Support and guidance, both pre and post implementation is part of our DNA. In other words, if you need help with our tech or have questions, we’re here to provide the answers.

At VISUA we understand that there are always going to be edge cases in which AI alone is not enough. Regardless of whether you are using our API or No-Code options, we can provide a full end-to-end service whereby we make mid-flow decisions to eliminate false positives or negatives.

We can also provide a full visual moderation service, following your rules and guidelines. This allows us to determine which content should be removed from your platform, drastically reducing the need for final manual determination by your team.

Yes, of course. We can give you a demo of how our visual content moderation API works. Just fill in the form at the bottom of this page or get in touch with us at [email protected] and we can arrange a time.