Case Study – Precise Product Color Identification from Images by Clickworker
Since 2016, clickworker has been supporting OTTO Versand with precise identification of product colors. Trained Clickworkers mark the dominant colors on images of clothing, accessories, furniture, and many other product types – up to three colors per product. The extracted color values are sent directly back to the client via API.
Get in touch with us! +1 (212) 878-6686 +49 201 95971830
Precise color recognition in product images is complex – especially with clothing, furniture, or accessories featuring color gradients, patterns, or shadows. Despite technological advances, AI still struggles to reliably determine the dominant color of a product visually.
That’s why OTTO continues to rely on human perception: only real people can evaluate visual impressions – such as the dominant color of a product – in all their nuance and context. At the same time, it was important to establish a scalable, automated process that ensures consistently high quality across millions of products.
clickworker implemented a quality-assured, scalable solution: Clickworkers were first tested through individual training with gold standard data. Only those who passed the test tasks were approved for the project.
OTTO’s product images were transmitted to clickworker via API. The Clickworkers selected up to three points with mouse clicks that represent the main colors of the product. If there was a significant deviation from the original color specification, a notification was automatically displayed – an additional quality check to prevent errors.
clickworker is happy to support you with qualitative categorization and color identification of your products – Request now!
Creation of video datasets to improve an eyetracking algorithm.
View Case Study
Fast and reliable assignment of products to online shop categories.
View Case Study
Relevance assessment of search results for an improved user experience in e-commerce.
View Case Study
