Retail Impact Stories: Automated quality assessment for images at scale with AI-powered image moderation

Hi?there,

Retailers face two big challenges when it comes to having product images on their site — one,?the quality of the image and the message it conveys to a shopper; two, the effort expended to ensure every single image adheres to site guidelines and brand values.?

Businesses have traditionally depended on manual processes to check and approve images for their site. Be it a brand selling its own products, or marketplaces where vendor onboarding is part of the process,?every image goes through a manual checking process. The problem is errors and biases, along with a slow go-to-market, result from manually checking every single image for all products across hundreds or thousands of sellers.?

Manual image moderation acquires complications when nuances like text and background have to be rigorously monitored, feedback loops need to be established and maintained, vendor relationships fostered based on image quality provided and team quality and efficiency need balance.?

Vue.ai’s conversations with retailers around the world helped us understand what retailers were looking for in a solution.?

1. A way to?accelerate the speed of onboarding images?while accurately checking if they match the website’s guidelines.?
2. Reduce the effort and stress on their teams?and establish a feedback loop that would save time, and money resulting in faster go-to-market.?

Here’s how Vue.ai automated quality assessment for images at scale with?AI-powered image moderation?for an e-commerce marketplace in Turkey.

The Customer

Hepsiburada is the only domestic and national player in the rapidly growing e-commerce market in Turkey.?They started using Vue.ai’s automated image moderation solution to help make the image QA process more efficient and effective. Vue.ai’s AI-powered image recognition technology automates image and content moderation. It can help ensure images are compliant with business guidelines - and can do this in half the time and in a way that would require minimal human intervention. AI also removes the prospect of human error - leading to faster and more accurate vendor image onboarding.

The Challenge

A manual process was in place to check for image quality. For the scale of operations that the marketplace was dealing with, manual image quality moderation was inadequate.?The customer had approximately?6 million images per month?with 20% of the overall volume going in for checks every month.

It took?10,000 man-hours roughly to check the fraction of the overall volume. The process was also resource-intensive, prone to errors that slowed down the time taken to monetize a product.?

Manual image quality detection threw up three critical issues:?

• Slow onboarding time
• High time and effort with manual quality checks
• Errors and inaccuracies in results

Vue.ai's image moderation?tool automatically checks the quality of images that sellers onboard into marketplaces. It checks for outliers and errors before the images go up on the site.?

The Solution

Two metrics were used to measure the quality of Vue.ai’s image moderation:

Accuracy %?

The Vue.ai solution would check the incoming images and provide a confirmation of images that met the guidelines set by the Hepsiburada team. Failed images were to be returned with information about the guideline that was not met. A higher accuracy percentage was indicative of how accurate the solution is in flagging images as accept and reject

Manual effort reduction?

The automated image moderation solution will help reduce the manual effort expended by the team to check every incoming image for quality adherence. The manpower saved here helps in focusing resources in areas where necessary.?

The Impact

Check out the entire case study here