Project Overview
The Company:
Retail AI development firm, California, USA
Industry:
Retail & Technology
Services:
Image Annotation
Business Needs
The client needed precisely annotated and varied image datasets to train their AI and ML models to manage inventories, recommend products, enhance product search experience and overall customer experience.
The Challenges
- Handling and annotating a vast collection of 1.2 million fashion and décor images within a tight deadline of 12 days.
- Upholding quality and achieving precision in annotations to train effective AI models.
- Understanding and categorizing a diverse and raw range of fashion products.
- Training the team to ensure project delivery on time without lowering annotation accuracy.
Solutions and Results
Want to annotate images at scale?
Speak to us todayHitech BPO created a customized solution with experts using sophisticated annotation tools:
- Trained a team of annotators on home décor, furniture, and fashion industry for domain knowledge.
- Developed a comprehensive workflow with defined SOPs and KPIs using proprietary tools for image annotation accuracy.
- Annotated a broad spectrum of fashion attributes, aligning with specific AI model requirements, using bounding box annotation.
- Implemented a detailed quality control process, using verification checklists to ensure data validity and accuracy.
- Organized the annotated images with unique IDs, further classifying them with keywords to support various AI model functions and simplify future identification and retrieval.
- Transformed annotated data into the client’s desired format for easy AI model integration.
- Completed the annotation of an extensive image dataset promptly, showcasing the team’s efficiency and project scalability.
Business Impact
96% increase in annotation productivity
1.2 Million Images annotated in 12 days
Improved model accuracy & performance
