How We Streamlined an AI-Driven Image Processing Background Removal and Color Correction Solution

Replace expensive third-party services for image processing

Shoob faced high costs and inefficiencies with capturing and processing student and employee photographs, storing them in Amazon S3. Their reliance on an expensive third-party service for background removal and color correction resulted in significant annual expenses to process 1 million images, despite many images going unpurchased. Shoob needed a cost-effective, in-house solution capable of handling both preview and production workloads with high accuracy and efficiency, all while reducing unnecessary costs.

Developed an AI-powered, in-house image processing pipeline

We developed a comprehensive in-house image processing pipeline to replace the Shoob reliance on external services. The pipeline efficiently handles production workloads, delivering high-quality background removal and color correction for all images. When a large image file is uploaded to S3, it automatically triggers the pipeline. The background removal process, hosted on Amazon SageMaker, uses advanced machine learning techniques to achieve precise edge removal, especially for complex features like hair, ensuring high accuracy.

Additionally, the pipeline leverages generative AI models, such as Claude 3.5 Sonnet on Amazon Bedrock, to provide color correction recommendations based on reference images. These recommendations are programmatically applied to ensure consistent and visually appealing results.

Efficiency, accuracy, and scalability

Significant Cost Reduction: Our solution has significantly reduced costs for the Shoob, cutting expenses by approximately 80%. Previously, they faced substantial weekly costs with their external image processing service for color correction and background removal.

Seamless Integration and Automation: The solution provides an API endpoint where users can submit an S3 URI. The system automatically processes the image for background removal and color correction, storing the final output directly in the S3 bucket, ensuring smooth integration into the Shoob's existing workflow.

Rapid Development and Delivery: Our team successfully designed and implemented this solution in just 6 weeks. This rapid delivery highlights our ability to quickly understand Shoob needs and deliver high-quality, effective solutions within a tight timeframe.

Transforming your image processing operations with cutting-edge AI technology is simpler than ever. As your trusted partner, our team will assess your unique requirements and craft a tailored solution using our expertise and proven methodologies.

Let's connect and explore the possibilities – contact us below.