Who: IPLAND and Azure AutoML
IPLAND, a provider of optimization solutions for major consumer goods companies, specializes in analyzing and fine-tuning product displays on store shelves. To meet the needs of its rapidly growing customer base, IPLAND sought a more scalable solution. By leveraging Microsoft Azure Automated Machine Learning (AutoML), IPLAND upgraded its Image Recognition (IR) service, resulting in improved image recognition speed, data collection, and a future-ready platform that can scale cost-effectively.
What: Modernizing retail environments with IR
IPLAND assists fast-moving consumer goods companies in optimizing their on-shelf presence. Using Image Recognition (IR) technology, IPLAND analyzes photos of store shelves to extract and surface critical performance data. This information helps brands fine-tune retail displays, improving product visibility and appeal to customers. By leveraging IR, IPLAND revolutionizes how brands monitor in-store inventory and optimize sales performance, providing real-time feedback and valuable insights for retail environments.
Before: Meeting challenges with scalability
As IPLAND’s customer base grew to include over a hundred major brands, its existing IR service infrastructure faced challenges in keeping up with the increasing demand. Speed and efficiency were crucial factors, as any delays in recognition could result in lower productivity and missed sales opportunities. IPLAND recognized the need for a scalable machine learning platform that could handle larger datasets and provide faster processing times.
The process: Implementing Azure AutoML
After careful evaluation, IPLAND chose Azure AutoML for its upgraded IR service. This decision was driven by Azure AutoML’s flexibility, ease of maintenance, and superior model management capabilities. IPLAND appreciated the cost-saving efficiencies and improved testing and training model management provided by Azure AutoML. The solution offered IPLAND the ability to choose the model architecture and hyperparameters, enabling further optimization and customization.
After: Setting new standards with improved recognition
Testing on the new IR service demonstrated recognition performance exceeding 95 percent accuracy. IPLAND achieved an average recognition speed of eight seconds per image in the field, surpassing their goal of 15 seconds by 33 percent. This performance improvement allowed IPLAND to scale and perform recognition in multiple streams simultaneously. The enhanced speed and accuracy of recognition were immediately appreciated by clients, leading to requests for service scaling.
Conclusion: Delivering results and driving business growth
The adoption of Azure AutoML has proven beneficial for IPLAND, enabling them to deliver superior results to their clients while optimizing costs and streamlining service administration. The improved IR service offers real-time adjustments for field workers, empowering retailers and brands to gain valuable insights into consumer behavior. IPLAND estimates lower total costs compared to their previous solution and continues to receive positive feedback from clients. With Azure AutoML, IPLAND remains committed to providing industry-leading alt text captions for millions of images, extending accessibility and supporting their goal of reaching a wider audience.
Leave a Reply
You must be logged in to post a comment.