Cali Intelligence was looking to reduce long retail checkout queues for major food retailers that cause lost sales, customer frustration, and reactive staffing decisions.
Using Ultralytics YOLO models, Cali Intelligence reduced retail checkout queues by 43% and improved staffing efficiency through real-time monitoring and alerts.
During peak hours, checkout lines can build up quickly in busy retail stores. As queues grow, wait times increase, staff become overwhelmed, and shoppers may abandon their carts before completing a purchase.
Most stores already have CCTV systems in place. However, these cameras are typically used only for surveillance and don’t provide real-time operational insights. This, in turn, means that store teams can’t detect congestion early or respond before queues become a problem.
Cali Intelligence tackles these operational challenges with AI-powered retail monitoring. By upgrading existing CCTV infrastructure with computer vision technology, they transform live video feeds into real-time operational data.
For instance, using Ultralytics YOLO models, their system can detect checkout lanes, identify active queues, and measure customer buildup. This helps store teams to respond quickly and prevent prolonged wait times.
Founded in 2020, Cali Intelligence develops AI solutions designed specifically for physical retail stores. The company was created with the goal of democratizing artificial intelligence in French retail and helping retailers improve performance and customer experience through computer vision.
A key challenge in physical retail is limited visibility into shop-floor activity. Unpredictable queues and uneven staff allocation make it difficult for store teams to respond quickly, especially during peak hours when checkout lines grow rapidly.
Retail teams are often forced into reactive decision-making rather than proactive management. Cali Intelligence addresses this gap by letting retailers better understand what is happening inside their stores in real time.
Over the past four years, Cali Intelligence has expanded its solutions across several retail sectors, including mass distribution, DIY, and ready-to-wear. Today, the company works with major French retailers such as Intermarché and Leclerc, supporting more efficient and responsive store operations.
Long checkout queues are one of the leading causes of shopping abandonment. A customer’s checkout experience often determines whether a sale is completed or abandoned.
Even when customers have filled their baskets, long queues can undermine purchase intent. This results in an immediate loss of sales.
In fact, the impact goes further than a single transaction. Repeated delays frustrate customers and may push them toward competitors that offer faster service. Over time, this erodes loyalty and reduces repeat visits.
Long queues also place significant pressure on store teams. At an operational level, management often struggles to respond quickly enough.
In many cases, teams react only after lines have already grown crowded, opening additional tills once the situation becomes urgent. This reactive approach forces staff into constant firefighting instead of enabling smooth, consistent service.
Staffing adds another layer of complexity. Without live queue data, it is difficult to know when and where additional support is truly needed. Often, stores end up overstaffed during quiet hours and understaffed during peak periods, leading to inefficiencies on both ends.
To improve store management and customer experience, Cali Intelligence automates checkout monitoring using computer vision through existing camera infrastructure. Their solution integrates directly with standard Video Management Systems (VMS), allowing store managers to receive instant alerts when queue thresholds are exceeded.
This enables teams to open additional tills or reposition staff before lines grow too long. At the centre of this solution are Ultralytics YOLO models.
Ultralytics YOLO models support key computer vision tasks such as object detection, which identifies customers in video frames, and object tracking, which follows those customers across frames over time. These capabilities make it possible for the system to monitor checkout areas, count customers, and identify emerging queues.
By detecting and tracking individuals in live video streams, the solution can also estimate wait times and flag developing bottlenecks. In particular, the system runs on compact, on-site servers using an edge-first architecture. This ensures round-the-clock operation while keeping customer data private.
In addition to real-time monitoring, the solution supports short-term forecasting. It can predict queue build-up up to 15 minutes in advance, helping managers align staffing levels with expected footfall.
Ultralytics YOLO models provide Cali Intelligence the ability to deliver high performance without the need for expensive cloud infrastructure. The models generalize well across different camera angles and lighting conditions, which supports rapid deployment across multiple stores with minimal retraining.
The Ultralytics YOLO models also support advanced object tracking. Instead of relying only on headcounts, the system can measure how long customers spend in line. This improves queue visibility and contributes to over 90% accuracy in real-world alert triggers.
On top of this, the YOLO-driven system is optimized to process even 3 to 6 camera streams at around 3 FPS per stream. This allows it to maintain detection accuracy while significantly reducing compute load, supporting efficient and scalable retail operations.
When Cali Intelligence deployed its Ultralytics YOLO powered solution across eight retail sites, the impact was both immediate and measurable. For example, at one site, the average queue length dropped from 7 to 4 customers, a 43% reduction within just two weeks.
Operational efficiency improved alongside customer satisfaction. During off-peak hours, the system reduced unnecessary checkout openings by up to 10%, enabling stores to align staffing levels more closely with actual demand and avoid wasted labor costs.
Meanwhile, detection performance remained stable across diverse store layouts and lighting conditions, maintaining a miss rate below 6%. High alert accuracy gave managers confidence to act quickly and make informed decisions on the shop floor.
The benefits also extended beyond real-time monitoring. Early testing of predictive labor optimization achieved a Mean Absolute Error (MAE) of 0.8, forecasting queue lengths within one customer of the actual count and enabling more proactive workforce planning.
Simply put, Cali Intelligence was able to leverage Ultralytics YOLO to convert in-store video into real-time operational intelligence, helping reduce wait times, optimize staffing, and enhance overall retail performance.
As Cali Intelligence continues to grow, the company plans to continue optimizing its edge performance using the Ultralytics Python package. The package provides a streamlined workflow for training, exporting, and deploying models, making it easier to implement performance improvements efficiently.
Building on this foundation, Cali Intelligence is exploring TensorRT and ONNX export formats to reduce inference time and improve hardware utilization on-site. The Cali Intelligence team is also evaluating a shift between Ultralytics YOLO model variants, moving from Medium to Small to improve efficiency while maintaining high detection accuracy.
Overall, Cali Intelligence is driving a shift in retail operations, moving stores from reactive management to proactive, data-driven performance.
Want to bring AI into operations? Visit our GitHub repository to learn more. Explore AI in logistics and computer vision in healthcare. Check out our licensing options to get started.
Ultralytics YOLO models are computer vision architectures developed to analyze visual data from images and video inputs. These models can be trained for tasks including Object detection, classification, pose estimation, tracking and instance segmentation.Ultralytics YOLO models include:
Ultralytics YOLO11 is the latest version of our Computer Vision models. Just like its previous versions, it supports all computer vision tasks that the Vision AI community has come to love about YOLOv8. The new YOLO11, however, comes with greater performance and accuracy, making it a powerful tool and the perfect ally for real-world industry challenges.
The model you choose to use depends on your specific project requirements. It's key to take into account factors like performance, accuracy, and deployment needs. Here's a quick overview:
Ultralytics YOLO repositories, such as YOLOv5 and YOLO11, are distributed under the AGPL-3.0 License by default. This OSI-approved license is designed for students, researchers, and enthusiasts, promoting open collaboration and requiring that any software using AGPL-3.0 components also be open-sourced. While this ensures transparency and fosters innovation, it may not align with commercial use cases.
If your project involves embedding Ultralytics software and AI models into commercial products or services and you wish to bypass the open-source requirements of AGPL-3.0, an Enterprise License is ideal.
Benefits of the Enterprise License include:
To ensure seamless integration and avoid AGPL-3.0 constraints, request an Ultralytics Enterprise License using the form provided. Our team will assist you in tailoring the license to your specific needs.
.webp)
.webp)
.webp)