success story

Warehouse management optimization achieved for a leading information management services company with generative AI

30x faster scanning and 80% accuracy of damage identification achieved

The information management services company has protected, secured, and preserved clients’ valuable data for over half a century. Whether it’s original documents and recordings of treasured artists, they are trusted with their most dear possessions.

Ensure that you open the links to enable the Next button, assuming you have read all the documents.
Okay & NextClick to Next
The Challenge

The information management services company's existing method for physical record identification, scanning, and replacement of damaged boxes in the warehouse was dependent on manual staff inspection. The labor-intensive procedure consumed precious time and resources, diverting employees from other essential duties.

After identifying the damaged boxes, the subsequent steps of requesting re-box orders were slow and cumbersome. The delay in re-boxing allowed the damaged items to continue in circulation, heightened the risk of further damage, and jeopardized product quality.

Relying on manual inspection, coupled with delayed re-boxing requests, not only hampers operational efficiency but also presents a notable threat of revenue loss.

The information management services organization set out to increase efficiency, reduce manual effort, and improve overall inventory management costs.

The Solution

Virtusa worked with the organization to develop a first-of-its-kind genAI-enabled drone detection product for rack optimization, personnel optimization, compliance, inventory control, and overall warehouse management optimization. 

The drones first mapped the site, acting as human inspectors. They flew the routes using an advanced AI-driven navigation system to capture images of bar codes, and LPNs and identify re-box candidates. Once the mapping was complete, each drone returned to its dock and securely uploaded the data. The drone footage is processed to extract frames, and the Segment Anything Model (SAM) from Meta is used to identify individual boxes from the frames.

The drone detection product uses Google’s transformer based BEiT vision model and Google’s Vision AI OCR model. The BEiT vision model has been trained to classify images of individual boxes as damaged, and Google’s Vision AI OCR model is used to extract barcode numbers from individual damaged boxes.

Replacement materials can be requested via the SKP API integration for re-boxing requests after the damaged boxes are identified.

Warehouse management optimization with generative AI
The Benefit
  • The generative AI-enabled systems helped the information management company achieve warehouse management optimization. The drone and artificial intelligence/machine learning (AI/ML) technologies reduced inventory management costs, increased the accuracy of detecting damaged boxes in their warehouses, and automated the tedious tasks of physical identification, scanning, and replacement.
  •  The system detects  subtle signs of damage that human inspectors might overlook with its advanced sensors and algorithms. This automation also decreases the  time between identifying a damaged box and initiating a re-boxing request. By reducing the reliance on manual inspections, our client freed up valuable employee time for warehouse staff to focus on more strategic tasks, thereby increasing overall productivity and efficiency.
  • Our initial PoC resulted in 30 times faster scanning than human inspection alone and saw an 80% accuracy of damage identification. The identification accuracy will increase as the model gets trained with more footage. 
Explore the power of generative AI

Enable business transformation with the latest AI technology

Learn more Click to Learn more
Related content