Online shopping is more popular than ever. Increasingly consumers are choosing to do the majority of their fashion shopping through websites and apps. Retailers have responded by making the online shopping process easier. To do this companies are turning to computer vision AI to create online fitting experiences where customers can try on items virtually.
Virtual wardrobes can match clothing items to the unique shape of customer’s bodies, and the same technology can be used for shoes and other apparel. Increasingly virtual fitting technology is being used to model accessories for online shoppers and this is the topic of today’s blog.
Firstly we will examine some of the specific use cases for virtual accessory fitting systems. Secondly we show how data annotation providers create training datasets for these applications. And finally we reveal how outsourcing annotation to dedicated services, like Keymakr, can save companies time and money.
AI use cases
Computer vision based AI models can recognise objects and place them correctly on images of the human body. As a result virtual accessory fitting apps are able to show customers how a particular item would look on them. This technology can support multiple product classes:
- Jewellery: High value items can be difficult to shop for online. Customers may find it hard to commit to pricey items, like jewellery, if they cannot view them and try them on in person. Virtual fitting experiences give online shoppers confidence when purchasing expensive products.
- Watches: It can be hard for online shoppers to gauge the size of items like watches. Virtual fitting experiences can show how a specific watch will fit on an individual's wrist. This important context improves the shopping experience and customer satisfaction.
- Glasses: Virtual fitting apps are helping customers find glasses and sunglasses that suit their features. This capability makes the online shopping experience much more personal and enjoyable.
Dataset creation for virtual fitting applications
The use cases shown above are made possible by image annotation. Annotators use tools to apply a range of labeling techniques to images of accessories. This allows AI models to recognise objects reliably.
- Polygon annotation: It is vital that training images accurately capture the shape of each clothing item. Polygon annotation allows annotators to plot points around an object and connect them with vertices. This gives annotators the freedom to precisely trace the outline of, for example, a necklace. This type annotation produces labeled items that adhere closely to their real world shape.
- Semantic segmentation: AI models must be able to identify and distinguish between different accessory objects in one image. Semantic segmentation assigns each pixel in an image to a class e.g. necklace, bracelet, wristwatch. This technique allows for a more granular understanding of a given image and improves the reliability of the virtual fitting experiences.
- Instance segmentation: Instance segmentation provides an additional level of detail for training images. This technique identifies each instance of each object appearing in an image, for example, each individual bracelet. This enables the virtual wardrobe to locate multiple, similar objects in one image.
The right annotation provider
Keymakr supports virtual fitting experience developers with high quality annotation services.
- Cost effective: It can be expensive for companies to create their own data annotation programme. Keymakr helps developers control annotation costs by offering affordable service without the burden of management responsibility.
- Scalable: Keymakr’s services can be flexibly scaled up and down. This allows developers to respond quickly to changing data needs.
- Annotation tools: Keymakr’s proprietary annotation technology has unique project management capabilities. Managers can assign annotation tasks based on worker performance to ensure that labels are always high-quality.
Contact a team member to book your personalized demo today.