A Chain Restaurant: SkyVision by DaoAI for Intelligent Supervision of Kitchen Behaviors in Transparent Kitchens

Chain restaurants with numerous stores across the country attach great importance to kitchen food safety supervision. Traditional supervision methods have many deficiencies, and the emergence of SkyVision by DaoAI provides a new way to solve this problem.

Scenario: The chain restaurant industry has developed rapidly in recent years. With the continuous increase in the number of stores, its scale has also grown significantly. These chain restaurants are usually distributed across the country, with hundreds or even more stores. To meet the requirements of transparent kitchens, cameras have been installed in the kitchens of each store. The main purpose of these cameras was originally to retrieve videos after problems occurred to find out the truth. However, during daily operations, the monitoring images captured by these cameras are mostly idle and not fully utilized. The headquarters lacks real-time understanding of the operational compliance of each store's kitchen and can only conduct supervision through irregular spot checks, which is inefficient and difficult to cover all stores comprehensively.

Pain Points: Why Is It Difficult?

In terms of the supervision scope, chain restaurants have a wide distribution and a large number of stores. Take a chain restaurant with hundreds of stores across the country as an example. Each store's kitchen needs to be supervised. It is impossible to achieve real-time and comprehensive supervision by manually staring at the monitoring wall to check the kitchen situation of each store. Assuming that it takes 1 minute for manual inspection of each store's monitoring image, it will take several hours to check all the monitoring images of hundreds of stores. Moreover, during this process, manual inspection is prone to fatigue and negligence, resulting in some violations not being detected in time.

In terms of the timeliness of detecting violations, traditional supervision methods often only discover problems in the kitchen when there are customer complaints or inspections by relevant departments. For example, problems such as not wearing gloves, hairnets/masks, someone smoking on the ground, and cross-operation of raw and cooked food may have existed in daily operations for some time. However, due to the lack of effective real-time supervision means, they have not been discovered. By the time the problems are exposed, they may have already had an adverse impact on the brand image and food safety.

In terms of supervision costs, manual screen-watching requires a large amount of labor costs. Moreover, due to the low efficiency of manual supervision, more manpower may be needed to ensure the coverage of supervision, which further increases the operating costs of the enterprise. In addition, post-event tracing requires a lot of time and energy to search for and analyze video materials, which also increases the supervision costs.

Technical Principle

The SkyVision system by DaoAI uses advanced video intelligent analysis algorithms. Based on deep learning technology, the system is trained with a large amount of kitchen behavior image and video data, enabling it to accurately identify various violations. The system directly accesses the existing cameras in the stores without the need to replace hardware. The cameras transmit the captured video data to the SkyVision system in real-time, and the system analyzes the video data in real-time. During the analysis process, the system extracts key features from the video, such as the appearance and action features of personnel, and compares these features with pre-set rules. If a feature does not match the rules, the system will determine it as a violation.

Compared with traditional manual supervision methods, the SkyVision system has obvious advantages. Traditional manual supervision mainly relies on human vision and attention, which is easily affected by factors such as fatigue and emotions and cannot achieve real-time and comprehensive supervision. The SkyVision system can monitor the kitchen 24 hours a day without fatigue or negligence. At the same time, the system has a high recognition accuracy and can quickly and accurately detect various violations, greatly improving the efficiency of supervision.

Typical Application Scenarios

• Detection of personnel protective equipment wearing: The system will detect in real-time whether employees are wearing gloves, hairnets, and masks. The difficulty lies in that the movements and postures of personnel may block the protective equipment, making it difficult for the system to accurately identify. For example, when a person raises their hand or lowers their head, the mask may be partially blocked. In this case, the system needs to use complex algorithms to determine whether the mask is worn correctly.
• Detection of smoking behavior: The system will identify whether there is anyone smoking in the kitchen. The difficulty is that the shape and color of smoke may be affected by environmental light and ventilation conditions, and some objects similar to smoke, such as steam, may be misjudged as smoking. The system needs to conduct in - depth analysis of the characteristics of smoke, such as the diffusion speed and color change of smoke, to accurately determine whether it is a smoking behavior.
• Detection of mobile phone use behavior: The system will identify whether employees are using mobile phones during working hours. The difficulty is that mobile phones are small in size and may be blocked by the body of personnel. The system needs to analyze the hand movements and concentration of personnel to determine whether they are using mobile phones.
• Detection of cross-operation of raw and cooked food: The system will identify whether there is cross-operation of raw and cooked food in the kitchen. The difficulty is that the appearance of raw and cooked food may be similar, and the position and state of food may change continuously during the operation. The system needs to comprehensively analyze the color, shape, and operation process of food to accurately determine whether there is cross-operation of raw and cooked food.

Implementation Case

There is a chain restaurant with hundreds of stores across the country that has been facing difficulties in kitchen supervision. Before the launch of the SkyVision system by DaoAI, the enterprise mainly relied on manual screen-watching to supervise the kitchen. The workload of manual screen-watching was huge, and the discovery rate of violations was low. On average, less than 10 violation events were discovered per month. At the same time, due to the inability to detect violations in real-time, some problems were only discovered during customer complaints or inspections, bringing certain risks to the enterprise's brand image and food safety.

During the launch of the SkyVision system, the DaoAI team directly accessed the existing cameras in the stores without the need to replace hardware. The entire launch process was fast and convenient. After the launch, the system immediately started to work. Through the real-time detection and alarm function of the system, violations in the stores can be identified and alarmed immediately when they occur. The number of violation events discovered per month increased to more than 30 on average, and the handling of violation events was advanced from “after the event” to “during the event”.

SkyVision turns record-keeping into real-time intervention, making the transparent kitchen a truly operable food safety defense line from a compliance decoration.

DaoAI's Solution and Product

The SkyVision system by DaoAI provides a complete intelligent supervision solution for the kitchen behaviors of chain restaurants. The system reuses existing cameras and can be launched without replacing hardware, greatly reducing the transformation cost. At the same time, it can be quickly promoted in many stores. The system can identify violations such as the wearing of gloves/hairnets/masks, smoking, and mobile phone use, and immediately push alarm information to store managers and the headquarters, realizing real-time alarm to individuals. All violation events will be automatically screenshot and recorded, and archived according to stores, time, and types, forming an auditable compliance ledger for the enterprise's subsequent management and tracing.

Quantitative Results

By using the SkyVision system by DaoAI, the chain restaurant has achieved remarkable results. In terms of detecting violation events, the detection rate of the system has reached over 98%, and the missed detection rate has been reduced to below 2%, greatly improving the discovery rate of violations. In terms of manual workload, the workload of manual screen-watching has decreased by over 80%, and the enterprise can allocate more manpower to other important tasks. In terms of violation event management, after continuous management, the number of violation events has decreased by over 70%, and the transparent kitchen has truly become the enterprise's food safety defense line.

This article was generated by AI. Customer cases are simulated scenarios based on real product capabilities and figures are illustrative; see product pages for official benchmarks.