One of the world’s largest sporting goods retailers with 1500+ stores across more than 45 countries needed its decision-making to be flexible and responsive to retain its leadership position. To be able to do this, it needed real-time analytics of store visitors, which is typically run using footfall data. However, the client wanted to link footfall data with POS data to improve store performance effectively and increase overall customer satisfaction.
After evaluating several vendors, the company chose Indium Software to implement a cognitive analytics solution leveraging computer vision to achieve its goal. Indium enabled this by using the existing security cameras:
- To generate heat-maps across the store
- Using ImageAI for facial recognition, identify customers and count the number of customers who enter the store at a given time period
- Incorporate a customized functionality to allow the cameras to count the number of people who performed a particular activity – like walk to a particular section of the store
The data collected from these cameras was used to analyze customer behavior at specific zones within the store. This was done by building comprehensive dashboards with real-time data refresh.
Indium built a neural network model by training various classes of the object or person, using 2000+ images for image processing and video analytics with maximum accuracy. By creating and testing annotations using sample videos, it was able to constantly improve the accuracy (over time, accuracy went up to 80%). Within the data points that were gathered, outliers were identified.
Since the visitor statistics was linked to the POS systems, they were able to track the conversion rate, improve product placement and implement cross-selling tactics. The solution had an easy-to-use interface that provided all stakeholders with better insight into customers’ behavior as well as shelf zone analysis.
As a result, the retailer was able to experience a 15% increase in customer satisfaction.
Deriving insights using Cognitive Analytics
According a report published by MarketsAndMarkets, the broader market for computer vision-enabled solutions is expected to grow from USD 15.9 billion in 2021 to USD 51.3 billion by 2026 at a CAGR of 26.3%. Of course, this includes the use of computer vision for factory inspections, audits, quality and safety management, etc.
Computer Vision, simply put, refers to enabling computers to identify and process objects in images and videos the way a human brain does. It helps computers to “see and understand” content in photographs and videos. By integrating AI capabilities to computer vision, we’re now able to conduct analytics on data from images and videos.
Today, it can be used to identify an item with 99% accuracy as against 50 percent a decade ago.
There are use cases for cognitive analytics and computer vision in a variety of industries such as:
- Energy, Power & Industrials: Computer vision can enhance safety, efficiency, and regulatory compliance of the power and energy industry by monitoring equipment for preventive maintenance and inspect linear assets such as power lines and pipelines for safety. It can also help with regulating danger zones and alerting in case an employee crosses a designated safety threshold.
- Manufacturing: Computer vision is being used in production lines for audits, inspections and overall quality management. Not only can it be used to detect defects, but data from other cameras and sensors can be cross-referenced to identify the root cause of problems. This can speed up repairs and prevent expensive downtime. It can also be used to prevent mislabeling and shipping errors. Workplace safety is another area where it can be used effectively.
- Retail: From improving customer satisfaction to inventory management, computer vision can be used in many ways to improve sales and profitability. Managing the store environment, enabling self-check-out, prevent shrinkages, and improve POS accuracy are some of the other ways in which it can be used.
- Transportation & Logistics: Right from autonomous vehicles to managing transportation and logistics, computer vision can be used to streamline operations and minimize supply chain disruptions. It can help improve the safety and efficiency of transportation fleets by ensuring proper docking, loading, fueling, and tire pressure.
- Healthcare: Medical diagnostics can benefit tremendously from computer vision risk assessment and early detection of disease. It can be used to ensure diligent conformance to safety practices such as handwashing among medical staff, manage inventory in pharmacies, ensure sufficient stocks in hospitals and clinics. It can also help in lowering administrative costs through automated document processing.
- Legal Enforcement: It can be very effective in preventing crimes by scanning live footage from public spaces and detecting weapons or identifying suspected behavioral patterns.
Indium’s Comprehensive Solution for Cognitive Analytics
Indium is a technology solutions company with deep expertise in digital, data engineering, data analytics services. With its global presence serving customers ranging from innovative product startups, Fortune 100 and global enterprises, Indium’s key differentiators are its specialization in:
- Ai, Advanced Analytics & Text Analytics CoE
- Big Data, Data Engineering, Stream Processing & Data Virtualization
- Low-code development across platforms
Indium Software has a team of cognitive analytics experts who work on the following areas:
- Image classification and analytics
- Object detection
- Facial recognition
- Video analytics
- Text analytics
- Speech recognition
We use FCN (Fully Convolutional Network), Mask-RCNN (Region-based CNN), and Detectron2 for image and video analytics; and deep neural networks and Optical Character Recognition (OCR) for text analytics and speech recognition.
Our range of solutions includes:
- Room Type Classification: Convolutional Neural Networks (CNN), YOLO, VGG16 Is used to classify room type using a pre-trained model. Based on their characteristics, images are categorized into different room types.
- Room Object Type Classification: Convolutional Neural Networks (CNN), YOLO is used to build pre-trained models to identify images and tag the object type in the room. A set of images is used to improve the accuracy of the model items in the image and room classified with a bounding box drawn around.
- Facial Recognition: Indium uses a sophisticated, scalable face recognition system, Multi-level CNN architecture, to detect, recognize, and analyze human faces in images. The system can also detect different face-related attributes such as head position, facial hair, emotion, gender, and age among others.
To know more about how Indium can help you build your cognitive analytics model integrating it with computer vision, contact us now.