Companies are increasingly using user-generated images and videos for engagement, but managing inappropriate content can be a challenge. Amazon Rekognition offers pre-trained and customizable AI capabilities for content moderation. With the new Custom Moderation feature, companies can enhance the accuracy of the moderation model and tailor it to their specific needs. The feature allows for training custom adapters with as few as 20 annotated images and provides improved performance and cost savings.
Enhance Content Moderation with Amazon Rekognition Custom Moderation
Companies often rely on user-generated images and videos for engagement, but ensuring that this content is consistent with policies and fosters a safe online community can be challenging. Traditional approaches like human moderators or reactive complaint responses are not scalable or efficient.
Amazon Rekognition offers a solution with its Content Moderation feature, which uses AI to detect inappropriate or unwanted content in images and videos. By using Rekognition, human moderators only need to review a small percentage of content, allowing companies to focus on more valuable activities while achieving comprehensive moderation coverage at a reduced cost.
Introducing Amazon Rekognition Custom Moderation
The Custom Moderation feature further enhances the accuracy of the pre-trained moderation model for business-specific data. With as few as 20 annotated images, companies can train a custom adapter in less than 1 hour, extending the capabilities of the moderation model for higher accuracy. The trained adapter can then be used with the existing DetectModerationLabels API operation to process images.
To create a custom moderation adapter, follow these steps:
- Create a project
- Upload the training data
- Assign ground truth labels to images
- Train the adapter
- Use the adapter
By following these steps, companies can tailor the Rekognition pre-trained moderation model for improved performance on their specific use case, without requiring machine learning expertise. The service is fully managed and offers a pay-per-use pricing model.
Best Practices for Training
To maximize the performance of the adapter, it is recommended to:
- Include sample images that capture representative errors to improve accuracy
- Provide true positives and true negatives in addition to error images
- Supply as many annotated images as possible for training
By following these best practices, companies can achieve optimal results with the custom moderation adapter.
Amazon Rekognition Custom Moderation is now generally available in all AWS Regions where Amazon Rekognition is available.
About the Authors:
Shipra Kanoria is a Principal Product Manager at AWS, with a focus on machine learning and artificial intelligence. Aakash Deep is a Software Development Engineering Manager based in Seattle, specializing in computer vision and AI. Lana Zhang is a Senior Solutions Architect at AWS, specializing in AI and ML for Content Moderation, Computer Vision, Natural Language Processing, and Generative AI.
If you want to leverage AI to enhance your company’s operations and stay competitive, consider Amazon Rekognition Custom Moderation. To explore more AI solutions and get advice on AI KPI management, contact us at hello@itinai.com or visit our website at itinai.com.
Evolve Your Company with AI
Discover how AI can redefine your way of work by identifying automation opportunities, defining KPIs, selecting the right AI solution, and implementing gradually. For AI KPI management advice, connect with us at hello@itinai.com. Stay updated on leveraging AI by following us on Telegram at t.me/itinainews or on Twitter @itinaicom.
Spotlight on a Practical AI Solution: AI Sales Bot
Explore the AI Sales Bot from itinai.com/aisalesbot, designed to automate customer engagement and manage interactions across all customer journey stages. Redefine your sales processes and customer engagement with AI. Explore solutions at itinai.com.
