Smoking Detection in Public: Build a Smoke-free Environment for Smart City
Hardware: reComputer J1020v2 with NVIDIA Jetson Nano
Application: Smoke Detection
Industry: Smart City
Deployment Location: Malaysia
Detecting smoking is important due to its multifaceted impact on health, safety, and the prevention of fire incidents. Smoking in public places not only poses health risks to both smokers and non-smokers in shopping malls, restaurants, and even school campuses but also introduces a significant safety hazard. The alarming correlation between smoking and fire incidents, especially in sensitive areas, emphasizes the critical need.
Recognizing this, many public infrastructures, such as airports, high-speed trains, gas stations, and flammable and explosive warehouses, are implementing measures to detect and control smoking behaviors. Equipping such areas with accurate and efficient smoking detection systems becomes crucial to timely identifying fire hazards. This not only safeguards the well-being of individuals but also ensures that firefighters and site management personnel can respond promptly to potential threats, minimizing the risk of fire-related incidents.
Challenge
Challenges may occur while building the detection system easily and efficiently. the traditional method could be the smoke detector which shows better performance for large-range coverage smoke and extended distance. However, the detection accuracy may be influenced by environmental conditions, such as weak lighting background and sparse smoke.
Now detection system with image processing and computer vision comes out as standard, but the limitation appears soon when dealing with small targets, such as identifying smoking behavior as a cigarette on the hand, resulting in issues like low detection accuracy and missed detection.
Solution
To address the challenges in small target detection, the two-stage learning approach can be implemented to enhance the accuracy and efficiency of the process. In the initial stage, we can leverage a hand detection model based on the Yolo algorithm to precisely locate the hand within the entire image. This strategic step significantly reduces the computational load for subsequent smoking detection.
Moving to the second stage, we can capitalize on the discriminant information inherent in the object itself. Here, a semantic segmentation model is employed to discern pixels within the identified smoke area, allowing for comprehensive utilization of surrounding pixel data. Specifically, the smoke semantic segmentation model, utilizing U-Net, determines the presence of smoke, culminating in the final decision regarding smoking behavior. Importantly, all these models undergo inference on the reComputer Jetson Nano with TensorRT for optimized acceleration.
By integrating these advanced models, it can realize real-time detection of smoking at the edge, meet the real-time requirements of intelligent monitoring and analysis scenes, and well reduce bandwidth consumption, protect data privacy, and can be combined with cloud storage to achieve real-time alarm and traceability.
Seeed NVIDIA Jetson Ecosystem
Seeed is an Elite partner for edge AI in the NVIDIA Partner Network. Explore more carrier boards, full system devices, customization services, use cases, and developer tools on Seeed’s NVIDIA Jetson ecosystem page.
Join the forefront of AI innovation with us! Harness the power of cutting-edge hardware and technology to revolutionize the deployment of machine learning in the real world across industries. Be a part of our mission to provide developers and enterprises with the best ML solutions available. Check out our success case study catalog to discover more edge AI possibilities!
Discover infinite computer vision application possibilities through our vision AI resource hub!
Take the first step and send us an email at [email protected] to become a part of this exciting journey!
Download our latest Jetson product Catalog to find one option that suits you well. If you can’t find the off-the-shelf Jetson hardware solution for your needs, please check out our customization services, and submit a new product inquiry to us at [email protected] for evaluation.
About Author
