FireNet: A Specialized Lightweight Fire & Smoke Detection Model for Real-Time IoT Applications (1905.11922v2)

Abstract: Fire disasters typically result in lot of loss to life and property. It is therefore imperative that precise, fast, and possibly portable solutions to detect fire be made readily available to the masses at reasonable prices. There have been several research attempts to design effective and appropriately priced fire detection systems with varying degrees of success. However, most of them demonstrate a trade-off between performance and model size (which decides the model's ability to be installed on portable devices). The work presented in this paper is an attempt to deal with both the performance and model size issues in one design. Toward that end, a `designed-from-scratch' neural network, named FireNet, is proposed which is worthy on both the counts: (i) it has better performance than existing counterparts, and (ii) it is lightweight enough to be deploy-able on embedded platforms like Raspberry Pi. Performance evaluations on a standard dataset, as well as our own newly introduced custom-compiled fire dataset, are extremely encouraging.

• The paper introduces FireNet as a shallow neural network with three convolutional and four dense layers, achieving 93.91% accuracy on real-time IoT platforms.
• The paper leverages a diverse custom dataset of fire and non-fire images to improve model generalization across varied scenarios.
• The paper integrates IoT capabilities with cloud services like Twilio and AWS S3 to enable remote, real-time fire alert notifications.

FireNet: A Lightweight Solution for Real-Time Fire and Smoke Detection in IoT Applications

The paper "FireNet: A Specialized Lightweight Fire Smoke Detection Model for Real-Time IoT Applications" presents a novel neural network, FireNet, designed to address the need for effective, cost-efficient fire detection systems suitable for real-time applications on embedded platforms such as Raspberry Pi. The authors highlight the limitations of traditional fire detection methods, including physical sensors, which often suffer from false positives and require noticeable fire development before activation. Leveraging advances in deep learning and computer vision, FireNet offers a visual-based approach to enhance accuracy and reliability.

Key Contributions

1. Neural Network Architecture: FireNet is a shallow network featuring three convolution layers and four dense layers, totaling 646,818 trainable parameters and an on-disk size of approximately 7.45 MB. This architecture enables the system to achieve an adequate balance between detection accuracy and computational efficiency, supporting a frame rate of over 24 fps on a Raspberry Pi 3B, which is comparable to real-time human visual perception.
2. Diverse Dataset Compilation: The authors counter the limitation of existing datasets by introducing a new, diverse training dataset consisting of fire and non-fire images obtained from multiple sources, including self-shot videos. This diverse dataset is crucial for training models to generalize well across varied scenarios encountered in practical applications.
3. IoT Integration: The proposed system incorporates Internet of Things (IoT) capabilities that enable remote notification and alert functionalities. Utilizing cloud services such as Twilio and AWS S3, the system provides MMS alerts with visual evidence in case of fire detection, facilitating real-time response.

Results and Discussion

The evaluation of FireNet involved a custom test dataset comprised of 19,094 fire and 6,747 non-fire frames, with the model achieving an accuracy of 93.91% on this challenging dataset. Performance metrics such as recall (94%) and precision (97%) demonstrate FireNet's robustness in differentiating fire from non-fire scenarios, outperforming existing methods limited by extensive computation and dataset specificity.

The paper contrasts FireNet with previous models based on well-known convolutional neural networks (CNNs) like VGG16 and ResNet50, which although effective, present practical challenges due to their size and complexity when deployed on cost-sensitive and resource-constrained devices such as embedded systems. FireNet's reduced model complexity directly addresses these deployment challenges while retaining high detection accuracy.

Implications and Future Work

The implications of FireNet’s development are significant for the advancement of fire and smoke detection technologies in the field of IoT. It promises affordability and portability, which can lead to widespread adoption across various infrastructures prone to fire hazards, such as residential buildings, warehouses, and industrial facilities. The potential for integration with other IoT systems introduces possibilities for implementing comprehensive safety solutions.

The authors acknowledge ongoing research to further refine FireNet’s robustness and extend its capability to learn from even more diverse datasets. Such endeavors would focus on enhancing the model's adaptability to different environmental conditions and improving its real-time response in various operational contexts.

In summary, FireNet exemplifies a pivotal step towards intelligent, lightweight fire detection systems suitable for real-time implementation in IoT applications, marking significant progress in bridging the gap between advanced deep learning methodologies and practical, cost-effective deployment solutions. The work sets a foundation for continued exploration into sophisticated, embedded AI systems applicable in a wide array of safety-critical environments.