AN AI ASSISTED PEDESTRIAN SAFETY DETECTION SYSTEM

Abstract

This system integrates Convolutional Neural Networks (CNN), Sensor Signal Networks (SNN), and GSM communication to enhance pedestrian safety at crosswalks. The CNN is used to process real-time video footage from cameras installed at crosswalks to detect pedestrians and predict their movement using machine learning algorithms. This sensor data enhances the accuracy and reliability of the detection, particularly in poor visibility conditions or complex traffic situations. When a pedestrian is detected, a geo-fence is established within a 350 to 500- meter radius of the crosswalk, and the GSM system sends real-time notifications to mobile devices or vehicles within that range. These alerts warn drivers to slow down and notify nearby pedestrians of potential risks. By combining these three technologies, the system creates a robust, scalable, and real-time solution to significantly reduce accidents at pedestrian crossings, improving road safety for all users.

Specification

Description:The present invention develops a sophisticated crosswalk detection system integrated with
machine learning algorithms, primarily Convolutional Neural Networks (CNNs), Siamese
Networks, and landmark detection models (100). The system initiates by booting up and
loading the program. The system employs a Convolutional Neural Network (CNN) via pre-
installed cameras to analyse pedestrian's recognition (101). These features are processed
through convolution, pooling, and fully connected layers to create specific crosswalk
representations. Then to match recognized pedestrians within the crosswalk (105), comparing
live crosswalk features with stored data to identify individuals with limited training data (103).
The system continuously tracks each and every pedestrian's presence in the crosswalk and
recognizes pedestrians with timestamps in the crosswalk, updating in real-time as long as the
camera is active. This process continues until the day is completed , Claims:1) A Smart crosswalk detection system (100) comprising a Convolutional Neural Network
(CNN) as claimed in claim 1 wherein the system uses pre-installed cameras that are
placed in the traffic signals to analyse pedestrian's recognition (103). These features
are processed through convolution, pooling, and fully connected layers to create
specific crosswalk representations;
2) A Smart crosswalk detection system (100) Utilizing a Siamese Neural Network (SNN)
as claimed in claim 1 wherein the system recognizes pedestrians who are crossing the
road, comparing live pedestrians to identify individuals who are crossing the road (106);
3) A Smart crosswalk detection system (100) operating continuously as claimed in claim
1 wherein the system processes images, recognizing pedestrians, and captures images
of pedestrians crossing the road in real-time as long as the camera remains active and
pedestrians are within its view (102). The crosswalk is continually updated based on
the recognition results, adding timestamps for pedestrians recognized in each image
(102). If a pedestrian is crossing the road (106), then the camera captures the image of
the pedestrians who are crossing the road, a message is sent to the mobile phones within
the range of 350meter to 500meter with location enabled in that particular phone (108),
else the camera is active so that it captures the pedestrians who want to cross the road;
and
4) A Smart crosswalk detection system (100) This would be used in geo-fencing system
to track the proximity of users (drivers and pedestrians) to a crosswalk (107). If a
pedestrian is detected, the system sends a notification to drivers within a 350-500-meter
radius to slow down (108). Similarly, pedestrians receive alerts when approaching a
crosswalk with oncoming traffic. The SSN could feed additional information, such as
the speed of nearby vehicles or adverse weather conditions (102). This data helps
personalize alerts, increasing urgency based on current conditions. Use of AI models to
predict whether a pedestrian is about to cross, giving drivers pre-emptive warnings to
slow down or stop even before the pedestrian reaches the crosswalk. Users could
configure the types of alerts they wish to receive (106).

Documents