O slideshow foi denunciado.
Seu SlideShare está sendo baixado. ×
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Anúncio
Summer Training Report
On
Object Tracking using Python
At
Ducat India Pvt Ltd.
Submitted in partial fulfillment of the req...
2
TABLE OF CONTENTS
PAGE NO.
1. CERTIFICATE………………………………………………………………………...3
2. CANDIDATE’S DECLARATION……………………………………..........
3
CERTIFICATE
As issued by the company
(Supervisor)
Anúncio
Anúncio
Anúncio
Anúncio
Carregando em…3
×

Confira estes a seguir

1 de 20 Anúncio

Mais Conteúdo rRelacionado

Diapositivos para si (20)

Semelhante a Obj report (20)

Anúncio

Mais recentes (20)

Obj report

  1. 1. Summer Training Report On Object Tracking using Python At Ducat India Pvt Ltd. Submitted in partial fulfillment of the requirement for the award of degree of B.tech (CSE) Manish Raghav Roll Number :1501010002 (2015-2019) K.R. Mangalam University (School of Engineering & Technology) Gurugram
  2. 2. 2 TABLE OF CONTENTS PAGE NO. 1. CERTIFICATE………………………………………………………………………...3 2. CANDIDATE’S DECLARATION……………………………………........................4 3. ACKNOWLEDGEMENT……………………………………………………..............5 4. ABSTRACT…………………………………………………………………………….6 5. INTRODUCTION……………………………………………………………………...7 Python NumPy OpenCV 5.2 TRACKING METHOD…………………………………………………………..10 5.2.1 Absolute Difference Method 5.2.2 Haar Cascade Classifier 6. ABOUT THE COMPANY……………………………………………………............12 7. PRACTICALTRAINNING………………………………………………….............14 7.1 SOURCE CODE…………………………………………………………………………….............15 7.2 SCREEN SHOTS…………………………………………………………………………................17 7.2.1 FIGURE OF PHONE 7.2.2 FIGURE OF PEN 8. CONCLUSION………………………………………………………………………..19 9. REFERENCES………………………………………………………………………..20
  3. 3. 3 CERTIFICATE As issued by the company (Supervisor)
  4. 4. 4 CANDIDATE’S DECLARATION I, Manish Raghav, bearing roll no 1501010002, a student of B.Tech(CSE) hereby declare that I own the full responsibility for the information, results etc. provided in this report titled “OBJECT TRACKING” submitted to K R Mangalam University, for the award of B.Tech(CSE) degree. I have taken care in all respect to honor the intellectual property right and have acknowledged the contribution of others for using them in academic purpose and further declare that in case of any violation of intellectual property right or copyright I, as a candidate, will be fully responsible for the same. My supervisor should not be held responsible for full or partial violation of copyright or intellectual property right. This is to certify that the above statement made by the candidate is correct to the best of my knowledge. Dated: (Candidate Signature)
  5. 5. 5 ACKNOWLEDGEMENT It’s a great pleasure to present this report of summer training in Ducat India Pvt Ltd in partial fulfillment of B.Tech (CSE) under K R Mangalam University. At the outset, I would like to express my immense gratitude to my Training guide (Mr Diwakar) guiding me right from the inception till the successful completion of the training. I would also like to thank my friends and all my group members for their help and cooperation throughout the training. I also thank my friends, who have helped me during this study. In addition I thank one and all who have been instrumental in helping me complete this project. I am extremely grateful and indebted to my parents and my siblings for being pillars of strength, for their unfailing moral support, and encouragement. I treasure their blessings and good wishes and dedicate this study to them. Name: Manish Raghav Roll No: 1501010002
  6. 6. 6 ABSTRACT This project is the output of our planning, schedule, programming skills and the hard work and this report reflects our steps taken at various level of programming skill, planning and schedule. Our project is about Object Tracking. It is a computer technology related to computer vision and image processing that deals with detecting instances of semantic objects of a certain class (such as humans, buildings, or cars) in digital images and videos. Well-researched domains of object tracking include object tracking. Object Tracking used in traffic to track vehicle also on playground it track players as much as possible. Now a days this features used in self driving cars. In this project we are able to track a object when placed in the front of the web camera. A coloured line indicates the movement of object when moved in the front of the web camera
  7. 7. 7 INTRODUCTION The detection of stationary or moving targets, and tracking them on a real-time video streams is a very important and challenging task in order to protect fields from enemies. The enemies can be a human, an animal or even an object. The field is secures by drones or stationary sticks which include detection and tracking system. Video surveillance is a very popular research topic in computer vision applications that continuously tries to detect, recognize and track the targets. Object detection compromises located objects in the frame of a video sequence. Every tracking method requires a detection method in every single frame. Object tracking is the process of following one or more objects that found on detection process using a camera. Background subtraction is the most common detection method used from simple object trackers. It is based on comparing two successive frames. Mean shift method and haar cascade classifier follows this. OpenCV libraries provide us these methods and have some RGB color detection algorithms. Object tracking process can be simply defined in three steps with an order and these steps are controlled by the computer autonomously. 1. Detection of interested object in video stream. 2. Tracking target object in every single frame. 3. Analyzing the tracking process to recognize the behavior. Human brain can easily understand the visual frames which seen by eyes, but computers cannot do it without any special methodology. If we compare human and computer, camera module is the eye of the computer and the computer vision is the way that how brain understand the image. Computer vision is a field that includes some methods, processing and analyzing techniques for understanding images.Also known as image analysis, scene analysis and image understanding.
  8. 8. 8 PYTHON Python is a high-level, interpreted, interactive and OOPS language. Python was designed to be highly readable, uses English keywords frequently whereas other languages use punctuation and it has fewer syntactical constructions than other languages.  Python is interpreted-This means that you can actually sit at a python prompt and interact with the interpreter directly to write your program.  Python is object oriented- it means that python support object oriented style or technique of programming.  Python is beginner's language- it is a great language for beginner programmers and support a development of wide range of application from simple text. HISTORY: Python is developed by Guido van Rossum in the late eighties and early nineties at the national research institute for mathematics and computer science in the Netherlands. Python is derived from many languages, including C, C++, java and many other scripting languages. Python is copyright .like Perl python source code is now available under GNU General Public License (GPL). Python is now maintained by a core development team at the institute, although Guido van Russom still holds the vital role in directing its progress NumPy: NumPy is the fundamental package for scientific computing with Python. It contains among other things:  a powerful N-dimensional array object  sophisticated (broadcasting) functions
  9. 9. 9  tools for integrating C/C++ and Fortran code  useful linear algebra, Fourier transform, and random number capabilities Besides its obvious scientific uses, NumPy can also be used as an efficient multi-dimensional container of generic data. Arbitrary data-types can be defined. This allows NumPy to seamlessly and speedily integrate with a wide variety of databases Open CV: OpenCV (Open Source Computer Vision Library) is released under a BSD license and hence it’s free for both academic and commercial use. It has C++, Python and Java interfaces and supports Windows, Linux, Mac OS, iOS and Android. OpenCV was designed for computational efficiency and with a strong focus on real-time applications. Written in optimized C/C++, the library can take advantage of multi-core processing. Enabled with OpenCL, it can take advantage of the hardware acceleration of the underlying heterogeneous compute platform. Adopted all around the world, OpenCV has more than 47 thousand people of user community and estimated number of downloads exceeding 14 million. Usage ranges from interactive art, to mines inspection, stitching maps on the web or through advanced robotics. . FEATURES: 1 Easy-to-learn: python has relatively few keywords simple structure,and a clearly defined Syntax. 2 Easy-to-read: python code is more clearly defined and visible to eyes. 3 Easy-to-maintain: python's success is that it is easy to maintain. 4 Portable: python can run on a wide variety of hardware platform and has same interface on all platforms. 5 Database: python provide interfaces to all major commercial databases.
  10. 10. 10 5.1 TRACKING METHODS 5.1.1 Absolute Difference Method This method can be used for detecting moving objects with respect to the background. It also called as “background subtraction”. The idea behind this method is, taking two different frames and subtracts each other. As a result, stationary objects will be filtered so just the moving objects will be detected. Approach steps: 1- Take the background frame at time t. 2- Take the difference between input frame and background frame. 3- Apply the threshold to get the foreground mask. This method is easy to implement. For that reason, it is widely used in most of simple object detection and tracking application. It also allows to continuous tracking. In comparison, it works slow and have low accuracy. Thus, it does not preferred in complex projects
  11. 11. 11 5.1.2 HaarCascadeClassifier Conventional object detection and tracking methods gives accurate results but they are very slow. In order to fast reaction required applications cannot use these methods. To solve this problem, Paul Viola and Micheal Jones developed haar cascade classifiers. It is also known as Viola and Jones object detection framework. The idea is to introducing the objects to the computer before detection and tracking. During the detection process, computer tries to find the introduced object, with an appropriate algorithm. Positive image: the image that contains the desired object. Negative object: the image that does not contain the desired object. In order to machine learning process, positive and negative images are introduced to the computer. Then computer recognize the desired object, to using features below. First step in this process, taking the positive images and scanning it with the adjusted feature frames. Second step is to count the pixels in black and white area separately. Lastly controlling the dark and light sections and have some target values. Haarfeatures These frames are called as “feature” or “weak classifier”. They called weak classifier because individually they have no meanings. Most of these weak classifiers found in one object multiple times. If computer observe most of these, it means there is desired object located. Computer recognize that the nose is lighter than cheeks and eyes. With using appropriate weak classifier, define the desired object. Haar cascade classifier is capable of detecting eye, nose, head, mouth, hair, etc. in face detection application. While detection weak classifier sizes are adjusted dynamically. Now it is time to using negative images. Negative images important for fast responses. Negative images are also scanned by the computer with the same algorithm. The result of it, computer knows what to do not want to detect, and this saves a lot of time. Haar cascade classifier’s speed secret can be easily explained with one word, integration. Instead of adding all the pixel variables separately, taking integral of it is much easier and faster. Fortunately, the computer vision software which named as OpenCV has a library of haar cascade classifier. This library is designed with state-of-art algorithm to be user friendly, time saver.
  12. 12. 12 ABOUT THE COMPANY Ducat is one of the nation’s largest CORPORATE EDUCATION and Consultancy providers. It is a completely Indian company which understands India’s training needs & how they affect our unique place in the global economy. Ducat’s Service package includes product quality testing, training, consultation, facilitation and hand holding for the quality initiatives.  Ducat has its corporate office in Noida (UP) with its branches in Delhi/NCR, Jaipur and Gwalior; and it provides quality services to over 20,000 professionals annually.  With over 10 years of experience in the industry, Ducat is recognized as a leading expert in the design, development and delivery of technical and soft skills training - from individual courses and seminars to certification programs and full-scale training solutions.  Ducat offers a full spectrum of vendor authorized technical, business skills, project management and application courses designed to suit every skill level, as well as the ability to consult directly with organizations to tailor made learning plans for any number of employees. Our clientele list is exhaustive which includes names like: Xansa, Birla Soft, Oracle, Perot Systems, SDG, CSC, Outworx, Nippon Audiotronix Ltd., HCL, Teri Group, ITC, Vishal Mega Mart and many more. We provide support in technologies like .Net, IBM Mainframe, Java, Software Quality Testing, ERP (SAP & Oracle Apps 11i), Data Warehousing, SAS, Embedded Systems, Linux etc.  We listen to your requirements and design solutions which lead to smooth running of your business, and are more efficient for business people. We develop high performance, high quality applications from scratch, or in some cases, enhance the functioning of existing systems by building custom modules. We deliver on time, the cost effective solutions for the corporate. We follow-up with full training and consistent support to ensure that our clients are satisfied. We are extremely versatile, with the resources and skills required to tackle any project with confidence, regardless of the level of complexity
  13. 13. 13 DUCAT’s Services: DUCAT is into three major segments: 1)Training (Project Based): - We provide project based training on various latest and upcoming technologies like J2EE, .NET 2008, Embedded System, ERP, IBM Mainframe, Data Warehousing, Linux, SAS, PHP ,Artificial Intelligence ,Machine Learning, Robotics etc. 1.1)Corporate Training: - At Ducat, we provide corporate education and custom-content training solutions to some of the India's largest private and public sector organizations. We serve our clients in various sectors such as information technology, education, professional services etc. We are corporate trainer for XANSA, HCL, Sky Workz, Outworx, and Green Ply, Unipatch Rubbers, SDG, Birla Soft, Perrot systems, Oracle and many other leading companies. 2) Development: - We have created custom based software and web based solutions for many different types of businesses. Some examples of the web based solutions developed by us are www.noidapolice.com, www.classicsofindia.com, www.amarsinghonline.com, www.bareillypolice.com, www.viewsonnews.in, www.akbari.in, KMVN, TEXPROCIAL, DMS PRO V.O1 (DUCAT), Helpdesk 3.01 for Globus Clothing (Tata Venture) & PMS (Procurement Management System (Dalmiya Construction). 3) Consultancy: - DUCAT is a global consulting and technology services company specializing in industry-specific solutions, strategic outsourcing and integration services. Our consultancy and manpower supply services are an added advantage for our students. .
  14. 14. 14 PRACTICAL TRAINING I have written a basic & simple code of object tracking in python using numpy, Open Cv module. Since python is platform independent so it works on windows, Linux, iOS. SOURCE CODE:
  15. 15. 15
  16. 16. 16 Press F5 for output. Below is the screen shot:
  17. 17. 17 SCREENSHOTS These images shows the output of my application Figure 1 1.1 This image indicates sensitivity of my camera towards the blue shades. 1.2 This is possible because of use of RGB component of blue shades in my program. 1.3 In this shows that if we place object in front of this webcam then where ever there is blue colour it will start tracking that object , a red colour circle is formed after that wherever we move phone then phone is tracked by webcam which is seen as green line as shown in figure 1
  18. 18. 18 Figure 2 2.1 Another screenshot of my application that indicates the tracking a pen which has a blue colour cap. 2.2 In this image I have used a pen which has blue colour cap as you can see where there is blue colour a red circle has formed which indicate that it has tracked blue colour and movement of object is shown by blue line.
  19. 19. 19 CONCLUSION Summary of the project: In this I evaluate my training and summarize the experience gained. I may discuss the production, the efficiency, the work environment, and the technical management in the company. The report is based on the knowledge, which I acquired during my 30 days of summer training. Real time object detection and tracking on video stream is a very crucial topic of surveillance systems in field applications. To easily accessible product, the project constructed as low cost project. In this project, several methods are presented. We implemented different detecting and tracking methods. Algorithms works well for detection and tracking. The results are good for starting. Haar cascade method gives the most useful results for our project. Limitation i) Limitations in the current project: Some advanced features are not available. ii) Future scope of the project: So Python has a good future in this world where AI and ML studies are taking leaps in their growth day by day.
  20. 20. 20 REFERENCES 1. MERL, "Object Tracking and Understanding," Mitsubishi Electric Research Laboratories, July 15, 2004 http://www.merl.com/projects/ObjectTracking 2. VPS Naidu and J.R. Rao, “Object Tracking using Image Registration and Kalman Filter”. 3. J. L. Barron, D. J. Fleet, and S. S. Beauchemin. Performance of optical flow techniques. INTERNATIONAL JOURNAL OF COMPUTER VISION, 12:43–77, 1994.

×