ASYNCHRONOUS SPATIO

TEMPORAL MEMORY
NETWORK FOR CONTINOUES
EVENT BASED OBJECT
DETECTION
GUIDE:Prof. R. ANGELIN PREETHI, M.Sc., M.Phil., Ph.D.
PRESENTED BY
S.SANGEETHA
DEPARTMENT OF COMPUTER SCIENCE
 ABSTRACT
• Detecting and tracking objects are among the most prevalent and challenging tasks that a
surveillance system has to accomplish in order to determine meaningful events and suspicious
activities, and automatically annotate and retrieve video content. Detection can be achieved by
building a representation of the scene, frame difference is the method we use for background
subtraction, frame differencing is the intensity dissimilarity between two frames assuming that
intensity change of a pixel apparently indicate something changing in the image, that is a moving
object. It simply subtracts the current image from the previous or a reference image. After the
detection, it is then the tracker’s task to perform find its correspondence in the following frames
while constructing object’s trajectory. Here we use Kalman filter to do the tracking of the
objects. Thus in this way we can be able to detect and track the movable objects in a video.
 INTRODUCTION
• A moving Object detection has wide range of applications Such as Automated video
surveillance, mobile robot navigation, robot vision, monitoring of traffic, military
reconnaissance, security systems, animation purpose and human-computer interaction etc.

• Now-a-days, moving object detection is a diverse topic in the computer vision research. Here in
this paper, the two techniques are implemented in MATLAB, and results were compared.

• Two-frame difference, the objective of the approach is to detect the moving objects from the
difference between the existing frame and the reference frame.

• The frame difference method is the method of motion detection.

• This method adopts pixel-based difference to find the moving object ..

 CON…
• The first frame is captured through the continuous video and converted into grey scale image
and then sequence of frames is captured at regular intervals, and converted into grey scale
images.

• The absolute difference is calculated between the consecutive frames and the difference image is
stored. The difference image is translated into binary images, by optimal threshold.

• The advantage of this technique, it can resist the interference of light to some extent when
compared with the basic background subtraction technique.

• The input RGB video is converted into grey scale because the context of computing comprises
only two intensity values of the colour image i.e. '0' & '1'.

• And also, basically background subtraction technique needs a stable background which is very
complicated to extract in real time applications.
BLOCK DIAGRAM
 MOTIVATION

• An important stream of research within a computer vision, which has gained a lot of importance
in the last few years, is the understanding of human activity from a video.

• The growing interest in human motion analysis is strongly motivated by recent improvements
in computer vision the availability of low cost hardware such as video cameras and a variety of
new promising applications such as personal identification and visual surveillances.

• The goal of motion detection is to recognize motion of objects found in the two given images.

• Moreover, finding objects motion can contribute to objects recognition.

 OBJECTIVES
• The main objective of the project is to recognize pixels belonging to the same object.

• However, the present research is based on the following assumptions:

• A well fixed camera –stability is key if you want to isolate motion.

• Stable light, no flickering .

• Contrasting background .

• High camera frame rate and resolution

• Image noise, due to a poor quality image source.

• Gradual variations of the lighting conditions in the scene .

• Small movements of non-static objects such as tree branches and bushes blowing in the wind.
 PROPOSED METHOD

• Frame Difference Method

Detection of moving object from a sequence of frames captured from a static camera is widely
performed by frame difference method. The objective of the approach is to detect the moving
objects from the difference between the existing frame and the reference frame. The frame
difference method is the common method of motion detection. This method adopts pixel-based
difference to find the moving object

• Object Tracking
1.Point tracking algorithm
2.Kalman filter
 DEEP LEARNING

• Deep Learning is a subset of machine learning.

• Neural network attempt to simulate the behaviour of the human
brain allowing it to learn from large amount of data.
• Deep learning drives many artificial intelligence app & service
that improve automation , performing analytical without human
intervention.
 CNN-(CONVOLUTIONAL NEURAL
NETWORK)

• CNN-(Convolutional Neural Network) is a kind of network

architecture for deep learning algorithms and is specifically used for
image recognition & task that involve the processing of pixel data.
• To identifying & recognizing objects.
 INPUT/OUTPUT

• INPUT Live camera

• PROCESSING Segmentation
• OUTPUT 0bjects
 MATLAB

• The name MATLAB stands for matrix laboratory.

• MATLAB is a high-performance language for technical computing.

• It integrates computation, visualization, and programming in an easy-to-use environment

where problems and solutions are expressed in familiar mathematical notation.

• MATLAB is an interactive system whose basic data element is an array that does not
require dimensioning.

• This allows you to solve many technical computing problems, especially those with
matrix and vector formulations, in a fraction of the time it would take to write a program
in a scalar noninteractive language such as C or Fortran.
 REQUIREMENTS

HARDWARE REQUIREMENTS

• HDD: >90GB
• PROCESSOR: >Pentium IV 2.4GHz
• SYSTEM TYPE: 32bit / 64 bit
• RAM: >2GB
• OS: WINDOWS 7/8/8.1/10

SOFTWARE REQUIREMENTS

• TOOL : MATLAB 2019

• TOOL BOX : Image Processing Tool Box
 APPLICATIONS

• Self driving cars .

• Activity Recognition .

• People counting .

• Ball Tracking in sports .

• Traffic Monitoring.
 YOLO
•YOLO is real-time object detection. It applies one neural network to the
complete image dividing the image into regions and predicts bounding boxes
and possibilities for every region.

•Predicted probabilities are the basis on which these bounding boxes are
weighted. A single neural network predicts bounding boxes and class
possibilities directly from full pictures in one evaluation. Since the full detection
pipeline is a single network, it can be optimized end-to-end directly on
detection performance.
 OUTPUT LAYER

The output layer in a CNN as mentioned previously is a fully connected layer,

where the input from the other layers is flattened and sent so as the transform
the output into the number of classes as desired by the network.
INPUT OUTPUT
 FUTURE SCOPE

• Future enhancement may include alerting the user

• By sending multimedia SMS or email
• capturing video
• Streaming it online
 REFERENCES
1] Real-time Detection and Tracking for Moving Objects Based on Computer Vision Method, 2017 2nd International

Conference on Control and Robotics Engineering

[2] Moving Detection Research of Background Frame Difference Based on Gaussian Model, Computer Science &

Service System (CSSS), 2012 International Conference

[3] Kinjal A. Joshi, Darshak G. Thakore, ”A Survey on Moving Object Detection and Tracking in Video Surveillance

System "International Journal of Soft Computing and Engineering (USC E) Volume-2, Issue-3, July 2012.

[4] RupaJiS.Rakibe, BharatiD.Patil, ”Background Subtraction Algorithm Based Human Motion Detection”,

Intelllational Journal of Scientific and Research Publications, May 2013

[5] JoshanAthanesious, Suresh, ”Systematic Survey on Object Tracking Methods in Video”, Intelllational Journal of
Advanced Research in Computer Engineering & Technology (UARCET) October 2012, 242-247.
THANK YOU