case study : predicting wear and damage in the logistics to reduce costs

A popular furniture and home furnishing manufacturer in Japan collaborates with multiple distributors for product sales. Right across the local territory, the products of home furnishing and furniture have garnered a huge popularity among the customers, building a trusted partnership for over 10 years with them. Customers are excited when new product designs arrive and it sees a rapid passing out of the products from their logistics. Customer wants to improve their logistics service so as to deliver more products to customer doors and reduce customer grievances.

Challenge:

Lately, customer has discovered a new hurdle in their logistics operations that has resulted in wear and damage to their product line. With time, the prevalence of such affairs becomes so relevant. Customer was unable to meet the customer demands when they asked for the specific brand of their choice. It is the core area for improvement.  They wanted a system to do on time prediction of wear and damage of their products so as to comply with the customer demands and reduce scopes for customer disappointments. On top of that, it was aimed at detecting defects or finding the root cause of the product design at an earlier stage.

Solution:

SynergyLabs AI powered Visual Inspection or automated quality testing procedure was the primary solution for the purpose.

We know it is critically important to check with the quality of product and service to exercise long-term service success. But with the visual quality inspection system, the purpose can be achieved to some length only. Embedded automatic technology, this process can detect defects under certain conditions. It carries out pixel wise comparison between a referred image of the product and a specific image of a certain product. This is a method that only works under active environmental preconditions. With this system, the image of the product should be aligned with the inspection tool under properly exhibited perfect lighting conditions all across the system. This allows for all types of defect identification in advance. But, if those conditions are not met, an automated quality testing may fail and does not result in good benefits for owners.

Our only approach towards overcoming this challenge was based upon computer vision and machine learning. We could say one-of-its-kind technique developed by us for customer is nothing better than quality assurance tool integrated with computer vision and machine learning or Artificial Intelligence.

The model was built with neural network structures that employed high end artificial and machine learning and deep learning algorithms.

How we worked on it

Our AI powered deep neural network tool was programmed with supervised learning algorithms containing visual imaging of the products. It could help the system to identify defects of the goods from different perspectives. Using this advanced AI powered quality assurance tool, the rate of accuracy can be enhanced and it can be used widely to inspect product defects from all angles.

That’s apart; we embedded different attributes such as imperfect surface orientation, lighting, and contamination of irregular background textures on the goods. Based on these criteria, our tool was able to focus only on defects in advance. This approach could ease the process of manual activities that failed to identify defects of the product on time and contributed to decrease in production growth. So, our AI enabled tool is much improved and automated to increase performance capability in relation to identifying rare and complicated product defects.  

Result:

Embedded with powerful cameras and video analytics tools, our AI-based quality inspection tool was uber smart to enable high accuracy results as compared to traditional quality assurance tool subjective to errors in defect recognition. We also observed our AI powered visual inspection tool that check with quality of goods in the manufacturing site increases efficiency and productivity by almost 20%. At the same time, our tool was likely to take less time to process data and find defects with increased level of accuracy. Overall, we found that our product integrated with deep learning algorithms was capable of improving defect detection by 45% as compared to manual techniques. We also found it reduced shipment costs of the logistics associated with damaged goods during shipment. Customer was satisfied with the result and decided to adopt our AI powered solution all across its manufacturing site.