Fashion Data Annotator

Summary

We are a fully-female founded Fashion-Tech start-up that’s on a mission to change the way people search and shop online forever. In order to do this we need to create a large library of images of fashion items which are annotated (marked-up). Specifically each item in an image needs to be visually examined and a decision made on the exact type of item it is, noting the category and other descriptors including (but not limited to): fit, colour, pattern, material, and silhouette.

Goal: We are creating a high-functioning team which can perform the necessary data annotation on our catalogue of images which achieves, or ideally exceeds, the velocity goal of annotating an object in 5 minutes (12 objects per hour) at a 98% accuracy rate after the training period.

Core Responsibilities

Review an image (picture) containing one or more fashion items including apparel and accessories, and identify and document specific attributes. These attributes are pre-defined and described in a guide. Additionally, each individual item within the image must be masked (segmented or outlined clearly). Tags describing the attributes (e.g., category, colour, fabric, brand, etc.) must then be assigned to each identified and masked object. Accuracy will be measured against the information provided in the guide. A specified rate of completion is also required and assumed to be no more than 5 minutes per item (e.g. top, skirt, coat) for the full set of attributes (category, colour, fabric, brand, etc).

Key Skills & Knowledge:

• Ability to understand a visual and written rule set which details the our Taxonomy*
• Ability to visually assess individual fashion apparel and accessory items and correctly identify the full set of attributes and capture them in the tool provided
• Ability to use a basic annotation tool (a simple visual tool in AWS-Amazon Web Services)
• Keen attention to detail to ensure each item is correctly annotated accurately using our Taxonomy and Guides

*Note: This taxonomy covers standardised fashion terms as well as custom rule sets about

how we are determining identification and categorisation of both specialised terms (e.g.

peplum skirt) and more subjective attributes (e.g. colour identification: dark blue vs royal blue)

Resources & Support:

We will provide training on the following:

• Our Annotation Guide
• Our Annotation Processes
• Annotation Tool (set up in AWS Environment)

Work Schedule

As the client is UK-based, you will be required to work in UK daytime:

• Monday to Friday
• 14:00 - 23:00 IST (08:30 am - 17:30 GMT)

Pay & Benefits - What you’ll get in return:

• Annual CTC: 5 to 7 lakhs
• Fully remote role