Introduction

This case study explores how AI-powered technologies were leveraged to automate data extraction from product images, reducing manual efforts while ensuring data reliability.

Problem Statement

Previously, data extraction from product images was performed manually, leading to:

• Time-consuming processes

• Susceptibility to human errors

• Inconsistencies and inefficiencies in data collection

• Compromised data accuracy and reliability

Solution

To address these challenges, a user-centered approach was implemented to streamline the data extraction process while leveraging AI technology for enhanced accuracy and efficiency. The solution focused on:

• Implementing Optical Character Recognition (OCR) and Natural Language Processing (NLP) to extract text accurately from images.

• Designing an intuitive interface where users can easily upload product images.

• Utilizing AI to categorize and organize extracted data into a structured, easy-to-review format, minimizing manual corrections.

• Incorporating AI-driven validation checkpoints to detect and highlight potential inaccuracies before finalizing extracted data.

• Ensuring a smooth workflow from image upload to data verification, reducing cognitive load and enhancing user efficiency.

• Presenting extracted data in a structured, easy-to-review format, minimizing manual corrections.

Research

User Research & Key Insights

To understand the pain points, we conducted surveys and interviews with stakeholders in the fragrance industry. Key findings included:

• Manual extraction was slow and prone to errors.

• Inconsistent data formats caused difficulties in analysis.

• Users required a seamless way to extract and structure product data efficiently.

Personas

Persona 1: Data Analyst

• Needs: Accurate and structured data for reporting and analysis.

• Pain Points: Manual extraction is tedious and error-prone.

Persona 2: Product Manager

• Needs: Quick access to product details for decision-making.

• Pain Points: Difficulty in extracting complete and reliable information from images.

Ideation & Conceptualization

User flow diagrams to visualize the automated process.

Design & Prototyping

Low-Fidelity Wireframes

High-Fidelity Designs

Designed an intuitive UI for image upload and data review.

Home Page

Insight Table

Usability Testing

Process & Methodology

• Conducted usability tests with fragrance industry professionals.

• Gathered feedback on interface usability and data accuracy.

• Iterated based on user suggestions to improve workflow efficiency.

• Feedback on the analytics screen highlighted the need for customizable analytics to enhance data visualization and insights.

Impact & Results

• Enhanced Efficiency : Reduced manual effort and accelerated processing.

• Improved Accuracy : AI-powered extraction minimized errors.

• User-Friendly Experience : Simplified data validation and review.

• Cost Savings : Reduced dependency on manual labor.

• Actionable Insights : Enabled better decision-making with structured data.

Challenges Faced

• Blurry Images: Addressed motion blur issues through advanced image processing.

• Inappropriate Content or Angles: Enhanced image recognition to filter out irrelevant data.

• Brand Name & Logo Recognition: Improved AI model to identify stylized fonts and small brand logos.

Future Enhancements

• Expand to other industries requiring automated data extraction.

• Improve AI models for greater accuracy.

• Enhance UI for an even smoother user experience.