Surveillance Device Object Detection with Depthwise Separable Convolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video surveillance systems face issues with high power consumption and false detections due to the need for continuous video recording and transmission to cloud servers for object detection, leading to increased power usage and delayed alerts.
Innovation Solution
A video surveillance method and system that integrates a deep neural network (DNN) model with depthwise separable convolution layers on the surveillance device for real-time object detection, allowing the device to switch between standby and operation modes based on detected objects, reducing unnecessary recordings and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video recording and transmission to cloud server is performed continuously, then object detection capability is improved, but power consumption increases
Solution Approach 1:
The system segments the object detection function by integrating a lightweight object detection model directly into the surveillance device, separating it from the cloud-based processing. This allows local inference for immediate detection while reducing the need for continuous video transmission, thereby lowering power consumption while maintaining detection capability.
Solution Approach 2:
The system performs preliminary action by continuously monitoring for motion events and pre-processing video frames locally before transmission. Motion detection triggers selective recording and transmission only when relevant events occur, avoiding continuous power-intensive operations while maintaining detection readiness.
2Measurement precision
If video transmission to cloud server is performed, then object detection accuracy is improved, but alert delay increases
Solution Approach 1:
The system segments detection tasks by handling urgent motion detection events locally with the integrated model, achieving immediate alerts. Non-critical or follow-up analysis is performed on the cloud server, dividing tasks by urgency and complexity to eliminate delays while maintaining accuracy for critical events.
Solution Approach 2:
The integrated object detection model acts as an intermediary between motion detection and cloud processing. It performs initial inference locally to provide immediate alerts, then selectively transmits relevant data to the cloud for enhanced analysis, mediating between speed and accuracy requirements.
3Speed
If motion-triggered surveillance is used, then real-time detection is improved, but false detection rate increases
Solution Approach 1:
The system replaces the simple motion-triggered mechanical detection system with an intelligent object detection model that analyzes video content semantically. This substitution enables real-time detection while filtering false alarms by recognizing actual objects and their categories, distinguishing between relevant motion and irrelevant disturbances.
Data Source
AI summary
A video surveillance method includes the steps of detecting, by a motion detector, an object motion, in the field of view of a surveillance device when the surveillance device is in a standby mode; generating, responsive to a positive detection to switch the surveillance device from the standby mode to an operation mode, one or more images of the moving object; determining, by processing the one or more images with a deep neural network (DNN) model of an object detector, whether the objects contained in the one or more images belong to a given categories, wherein the DNN model comprises N (N is a positive integer and ranged from 4-12) depthwise separable convolution layers; and video recording, responsive to a positive determination, the moving object in the field of view of the surveillance device.