Object Detection & Tracking
- 1. Guided By - Prof. Shivprasad Patil, Head Of the Dept. Information Technology, NBN Sinhgad School of Engineering. Made By – Akshay Gujarathi 23 Vipul Oswal 47 Priya Adwani 53 Kadambari Metri 82
- 2. Introduction The modern world is enclosed with gigantic masses of digital visual information. To analyze and understand this huge sea of visual information, there exist many image analysis techniques. Those methods that automatically recognize and detect the objects prove to be of great use and provide a significant help in modern applications and devices. The semantic and syntactic contents of the images and videos can be recognized and further processed to get the necessary information.The potential uses of the image can be identified. The important content of image is the objects in the image. There exists a significant and essential need for object recognition techniques. Recognition is an important task in image processing and computer vision. A set of known tags can be used to identify what really the object is and help to extract information.
- 3. Motivation And Purpose The basic motivation behind this topic is that it is something that will overdo all the physical tasks. Robotics and smart systems are buzzing around all over the world. Object recognition and tracking reduces human efforts and provides efficiency. It is of interest as it may help humans to be aware of minute information about particular objects and reduce human tasks. Automatic recognition and extraction adds to the smart systems used today.
- 4. I. Object Representation: In a tracking, an object can be defined as anything that is of interest. For example, boats on the sea, fish inside an aquarium, vehicles on a road, planes in the air. People walking on the road are a set of objects that may be important to track in a specific domain. The appearance and shapes can be represented by object. First we will describe the representation of object shape. Representation of objects is very important in object detection and tracking. There are various ways used to represent objects.
- 5. Points: The figures to the right show the use of points in object representation. Primitive Geometric Shapes: Shapes like rectangles, ellipses can be used to represent objects. Object Silhouette And Contour: Contour representation defines the boundary of an object. The region inside the contour is called the silhouette of the object.
- 6. Articulated Shape models: Articulated objects are composed of body parts that are held together with joints. For example, the human body is an articulated object with torso, legs, hands, head, and feel connected by joints. Skeletal Models: Object skeleton can be extracted by applying medial axis transform to the object silhouette .This model is commonly used as a shape representation for recognizing objects. Probability Densities Of Object Appearance: The probability density estimates of the object appearance can either be parametric, such as Gaussian and a mixture of Gaussians, Parzen windows and histograms.
- 7. II. Difficulties And Problems In Object Detection. Illumination The lightning conditions may differ during the course of the day. Also the weather conditions may affect the lighting in an image Positioning The change in position must not affect the recognition system. Rotation The image can be in rotated form. The system must be capable to handle such difficulty
- 8. Mirroring The mirrored image of any object must be recognized by the object recognition system. Occlusion The condition when object in an image is not completely visible is referred as occlusion. Scale Changes in the size must not affect the Occluded car recognition system
- 9. III. Techniques for object recognition. Template Matching: Template matching is a technique for finding small parts of an image which match a template image. It is a straightforward process.
- 10. Colour Based: The object detection using colours involved in the objects is also significantly used and provide a simple to implement method. It provides potent information for object recognition. Color histograms prove to be simple and efficient and provide an edge for the same. The use and importance of color attributes for identifying objects has been proposed to us by Fahad Khan. This information has been segmented into two approaches which is the part based approach and the efficient sub-window approach. Feature combination, photometric invariance and compactness are the three major features that need to be taken into account while integrating or appending the color attributes with the object detection. Shape Based: Lately, shape has proved to be of great importance in object recognition. They have been explored dramatically to recognize objects in real world acquainted images. These features also provide an upper hand over local features like SIFT as most of the objects are illustrated and described by their shapes and textures such as different animals and other varying objects. They are most likely used to add an advantage to the local features.
- 11. IV. Extraction Of Object Background Subtraction: The background subtraction method is the common method of motion detection. It is a technology that uses the difference of the current image and the background image to detect the motion region, and is generally able to provide data included in object information. The background image is subtracted from the current frame. If the pixel difference is greater than the set threshold value T, then it determines that the pixels from the moving object, otherwise, as the background pixels.
- 12. Background Subtraction The result of image sequences computed by the method here is in the following figures. • When there is no movement in the frames. When there is no movement in the image sequences then the difference between the two images shows a black binary output image shows there is no difference in a single pixel. (a) Input first frame (b) Input second frame Binary image of difference image.Difference between two frames
- 13. • When there is movement in the frame. When there is movement in the scenes then the binary image of the difference between the two frames shows motion having white colour and where there is no change shows black colour. (a) Input first frame (b) Input second frame Binary image of difference image.Difference between two frames showing moving object
- 14. V. Multiple Object Detection & Single Object Detection Multiple Object Detection: An image may consist of many objects. It may comprise of a single object or they may be many. There exist different methods for the recognition of the same and these methods are independent of each other. The flowchart to the right depicts a simple overview of how the multiple objects can be detected. It involves a very simple procedure of training the detectors and then these detectors are used for identifying the objects either by extracting the features or the boundaries of the objects. These detectors need to be already trained for the different objects that exist and they work in efficient way to serve the purpose. An input image is tested against the detectors and compared and finally the output that is the final objects that are detected are displayed. Input ImageInput Image Object 1 Detector Object 2 Detector Object 3 Detector Training with Object 1 Training with Object 2 Training with Object 3 Output Image with all Objects detected.
- 15. Stationary Object Detection The flowchart provides an efficient and simple to implement procedure for stationary object detection. This process is simple and straight forward. The input is firstly segmented and then classified as either multiple or stationary using the time parameter and the other mathematical analysis. N Y Video Sequence Absolute Differencing Threshold Segmentation Multiple Object Tracking Background Modelling using Median Running Average Background updation If any Object Stationary for 1 second? Declared as Stationary
- 16. VI. Machine Learning Process In Object Detection. Each of the methods that have been reviewed and analyzed require machine learning to be an integral part of it since no matter what the Trained image is, the detectors have to be trained for the objects to be recognized and to do this the machine needs to be trained. So, this brings up the concept about Artificial Intelligence in terms of object recognition. The detectors basically keep on building their database by feature extraction or other attributes like color, shape and then these features are used to match with the objects in the input image. Input image Detected Object Detectors
- 17. VII. Object Tracking Object Tracking is a major phase involved in many remote sensing applications. Object tracking is to track an object (or multiple objects) over a sequence of images. It can be defined as a process of segmenting an object of interest from a video scene and keeping track of its motion, occlusion, orientation etc in order to extract the useful information. Point Tracking: Objects detected in consecutive frames are represented by points. Kernel Tracking: Kernel tracking is usually performed by locating the moving object, which is represented by an embryonic object region, from one frame to the next. Silhouette Tracking: In this approach Silhouette is extracted from detected object. Silhouette tracking methods make use of the information stored inside the object region.
- 18. VIII. Applications 1. Biometric recognition 2. Surveillance 3. Industrial inspection 4. Content-based image retrieval (CBIR) 5. Robotics 6. Medical analysis 7. Lane Detection
- 19. Introduction to Lane Detection What is Lane Detection? Technically Lane detection is defined, as a well-researched area of computer vision with applications in autonomous vehicles and driver support systems. The lane detection task involves understanding the topology of the lanes around the car.
- 20. Lane Detection System(LDS) Block Diagram Indicating the result by means of visuals or audio Detection of lane Input image from camera
- 21. Lane Detection Implementation Input image Selected lane and position Indicator on screen 1. A novel technique is used to recognize lane for a various road and illumination, lane markings conditions such as damaged road surfaces blocked by a car, shadow, backlights, etc. 2. The basic transform used will be HOUGH transform along with the segmentation of image concept to detect the lanes without any errors or flaws.
- 22. Lane Detection System Flow & Pseudocode
- 23. Overview
- 24. Applications Vehicle Driver Assistance Systems. Automated Surveillance. Military Applications. Security.
- 25. IX. Conclusion In this presentation, we have overviewed the following points – 1. Basic concept of Object Detection and Tracking. 2. Problems and difficulties in Object Recognition. 3. Representation of objects. 4. Techniques in object recognition. 5. Multiple and single object detection and machine learning process. 6. Object tracking. 7. Applications. Thus we conclude – • Object detection is a task of extracting Objects from specific frames/images. • Object detection is one of the most widely used concept in the field of Artificial Intelligence. • Has a great scope in future for the development of the modern world.
- 26. References & Bibliography Himani Parekh , Darshan Thakore , Udesang Jaliya, “A Survey On Object Detection And Tracking Methods”, International Journal of Innovative Research in Computer and Communication Engineering, February 2014. Kinjal Joshi , Darshak Thakore, “A Survey Of Moving Object Detection And Tracking In Video Surveillance Systems”, International Journal of Soft Computing And Engineering, July 2012. Sukriti Srivastava, Ritika Singal, Manisha Lumb, “Efficient Lane Detection Algorithm using Different Filtering Techniques”, International Journal of Computer Applications, February 2014.
- 27. Any Questions ?
- 28. Developed By- Mr.Akshay Gujarathi (Designing Head) Contact - 922 6666 86