EMOTION RECOGNITION USING SPEECH
PROJECT GROUP: BIJAL GAJJAR, NIDHI DESAI, RAJIV MISTRY
GUIDE: DR.S.T. GANDHE
Abstract: Speech based emotion classification method is represented. Five basic human emotions including anger, boredom, happiness, neutral and sadness are investigated. The methods include dct coefficients, amplitude classification, neural network based recognition and real time based recognizer. For neural network based recognition we got 90% average accuracy and in real time based recognizer we achieved 80% average accuracy.
 DSP BASED GESTURE RECOGNITION SYSTEM
PROJECT GROUP: PRANIT KALANTRI, SHRADDHA SHAH
GUIDE: DR.S.T. GANDHE
Abstract: Gesture Recognition lies at the core of augmented reality based devices. This project tries to coalesce two disparate aspects of engineering which are image processing and embedded systems. The project aims at developing and algorithm for recognizing static gestures using deterministic techniques and implementing these algorithms on an embedded platform. The first part of the project deals with the development of the algorithm. The underlying principle of the gesture recognition is singular value decomposition and normalized distances.
These concepts are used in comparison and classification of query gestures with database gestures. Principal Component Analysis is used for dimensionality reduction and ease of calculations. Image processing techniques are used prior to computation to achieve background, illumination and scale invariant images. The database comprises of such preprocessed images stored with varying orientations. The other part of the project deals with setup of an embedded platform for the compilation and implementation of the algorithms. And DSP based board is configured to interface a camera and input devices. Software setup is done to support graphics and Open CV. Compilation of the programs written from the algorithms is carried out on the board. Thus, the project aims at standalone implementation of gesture recognition which sets the foundation for portable augmented reality.
 ARM BASED GESTURE RECOGNITION SYSTEM
PROJECT GROUP: RISHKUL KULKARNI, ARJUN IYER, ABHISHEK JALGAONKAR
GUIDE: DR.S.T. GANDHE
Abstract: Gesture Recognition lies at the core of augmented reality based devices. This project tries to coalesce two disparate aspects of engineering which are image processing and embedded systems. The project aims at developing an algorithm for recognizing static gestures using deterministic techniques and implementing these algorithms on an embedded platform. The first part of the project deals with the development of the algorithm. The underlying principle of gesture recognition is singular value decomposition and normalized distances. These concepts are used in comparison and classification of query gestures with database gestures.
Principal Component Analysis is used for dimensionality reduction and ease of calculations. Image processing techniques are used prior to computation to achieve background, illumination and scale invariant images. The database comprises of such preprocessed images stored with varying orientations. The other part of the project deals with setup of an embedded platform for the compilation and implementation of the algorithms. An ARM11 based board is configured to interface a camera and input devices. Software setup is done to support graphics and Open CV. Compilation of the programs written based on the algorithms is carried out on the board. Thus, the project aims at standalone implementation of gesture recognition which sets the foundation for portable augmented reality.
 CURSIVE HANDWRITING SEGMENTATION AND RECOGNITION
PROJECT GROUP: PAYAL SHAH, ABHINEET SARKAR, PRAMOD MISHRA, RAVIKIRAN NANDE
GUIDE: DR.S.T. GANDHE
Abstract: The off-line cursive handwritten word recognition is a software algorithm implemented using neural network tool box in matlab. With the help of this algorithm the input handwritten image is segmented into characters. In the task of handwriting recognition, the letters which are interconnected are separated using histogram based image segmentation technique. For this, we need to detect the top end and bottom end of the character. Each letter is thus saved as an image and then given to the neural network, which is trained with sufficient number of pairs of input and output samples of all the 26 characters a training set is made. The algorithm used for neural network training is the back propagation algorithm. The final output of the system is the recognized text which can be further processed on a word application.
 AUTOMATED DRIVER FATIGUE DETECTION SYSTEM
PROJECT GROUP: APEKSHA SHENOY, KIRTI KHOPKAR, RUCHIR KANAKIA
GUIDE: PROF.K.T. TALELE
Abstract: A vision-based real-time driver fatigue detection system is proposed for increase in driver safety. Driver fatigue is evaluated on the basis of two parameters: rate of eye closure and yawning. Images of the driver are taken through a camera. Each individual image is checked for the face, eyes and the mouth. If they do not comply with the rules set for fatigue detection then, there is an immediate warning given in the form of an alarm. The system is tested on a hardware platform consisting of a TI OMAP 3530 chip along with other supporting peripherals. The experimental results obtained so far show great promise in the near future. The system can reach 20 frames per second for eye tracking. Efficiency of the system is near 90% in standard light conditions. However, the system may give erroneous results when the illumination level goes low.
 OBJECT CLASS RECOGNITION
PROJECT GROUP: SAIKALASH SHETTY, SANTOSH TELANG
GUIDE: PROF.K. T. TALELE
Abstract: We present a method to aggregate certain features in a particular class/group of objects into sort of a dictionary. Then they are compared to the image given by the user to be recognized. The recognition of an object requires detection first. The invariant features once extracted become the reference for the input image detection. After this, the image descriptor is used for classification and then consequent recognition. Stable local feature detection and representation is a fundamental component of many image registration and object recognition algorithms. The SIFT algorithm as being the most resistant to common image deformations. A general trainable system for object detection in static images has been proposed that examines (and improves upon) the local image descriptor used by SIFT.
Like IFT, our descriptors encode the salient aspects of the image gradient in the feature points neighborhood; however, instead of using SIFT smoothed weighted histograms, we apply Principal Components Analysis (PCA) to the normalized gradient patch. The PCA based local descriptors are more distinctive, more robust to image deformations, and more compact than the standard SIFT representation. Normalization and scaling is inherently done. This system can be applied to face, people, car, objects like cycles, bikes for road systems; and aero planes, ammunition, for security; retrieval for image mining of databases, etc.
 SPEECH RECOGNITION BASED MOTION TRACKER
PROJECT GROUP: PRASHANTH NAYAK, DIPAL MODI, RAJU PATEL
Abstract: The objective of the project is to design a speech operated motion tracker for detection of motion. Our tracker is an essential requirement of any surveillance system. Nowadays CCTV cameras are used for security purposes. These cameras can even track the object that is in motion but as they are fixed at a particular place. In this work we present an operational computer vision system for real-time detection and tracking of object in motion. The system captures video of a scene and identifies moving objects using wavelet segmentations, operated by Speaker through speech. We have implemented speech recognition using Wavelet Decomposed LPC which gives 80 percent accuracy. Using Wavelet segmentation we were able to detect motion even in varying brightness, this gives an advantage of tracking motion in the regions of low brightness. Motion tracker can be used in surveillance areas where the CCTV cameras fail, Motion tracker will be able to follow the object until a clear image of the object is retrieved.
 REAL TIME DATA ACQUISITION FOR BIOMEDICAL APPLICATIONS
PROJECT GROUP: SANKET MORE, SATISH PARATE, HARSHALA MASUREKAR
Abstract: Biomedical Engineering is the application of engineering principles and techniques to the Medical field. This field seeks to close the gap between engineering and medicine. The
Biomedical devices available today only handle single discrete parameter at a time. They do not measure blood pressure, body temperature, pulse rate etc. at a time. Hence they are efficient in measuring single parameter but not ultimate enough to deal with many body parameters. Our project on ‘Real Time Data Acquisition System for Biomedical Application’ deals with the biomedical parameters determined from blood, heart pulses and temperature. It senses these parameters and analyzes it mathematically to have logical and reliable output at the display. It is a multichannel system which eliminates the need of using discrete devices for measuring biomedical parameters. Every channel senses different parameter and gives the corresponding result to the display. The microcontroller decides which channel to select for the output at display in automatic mode and manual mode. It gives preference to the user to select which parameter to be displayed on the LCD. The system also has alarming facility to inform user about the serious need of medicine intake, if the parameter exceeds the range or goes below average. The system can be used as a personal check-up kit and continuous monitoring system in ICU. The system would also beneficial in fields of sports, fitness clubs, and military and other security forces.
 RAPID PROTOTYPING OF ELECTRICAL LOADS AND MACHINES ON AN EMBEDDED PROCESSOR
PROJECT GROUP: MANJIREE A. INGOLE, SWAMIT S. TANNU, AMRUTA J. VIDWANS
GUIDE: DR.Y.S. RAO
Abstract: Aim of this project is to develop a test bench which can be utilized to prototype electrical loads and machines. Induction heating system and Variable speed windmill were two case studies prototyped on the TMS320F28335 processor. In case of Induction heating system load and control system has been modeled and implemented in discrete time on the processor.
Current tracking and zero voltage switching of series resonant inverter were used as basic building block of this algorithm. Latency of the control algorithm is adjusted by the efficient resource allocation and optimizing the algorithm.
Complete system of variable speed windmill was modeled and implemented on the DSP processor. Flux control method was implemented for rectifier and inverter control. Each component of model was defined as task and adaptive task scheduling was used to minimize the latency.
 NON INVASIVE HEALTH MEASRURING SYSTEM
PROJECT GROUP: RUTA BELWALKAR, SHRUTI BHIDE, SARANSH MAKWANA
GUIDE: DR.Y.S. RAO
Abstract: Our project on ‘Non Invasive Health Measuring System’ deals with the biomedical parameters like blood oxygen saturation and heart rate. It senses these parameters non-invasively and analyzes it mathematically to have logical and reliable output at the display. It is a single input system which is capable of calculating these parameters eliminating the need of using discrete devices for measuring these biomedical parameters. This application used Ultralow-Power Consumption MSP430 mixed signal processor which is used to develop a single chip portable biomedical system. The system has application in future as personal check-up kit and continuous monitoring system in ICU.
 NEAR INFRARED REGION SPECTROSCOPY (NIRS) BASED TISSUE OXYGENATION MEASUREMENT
PROJECT GROUP: KAUSHIK GURAV, APURVA LIMAYE, ISHAAN KAUL
GUIDE: PROF. M.M. PARMAR
Abstract: Based on high speed ARM processor (ARM 7), a near infrared spectrum based processing and display system was developed. As an innovative method of tissue oxygenation measurement with the NIR spectrum, the system could compute real-time values of various tissue parameters especially oxy-hemoglobin and de-oxyhemoglobin. With the use of embedded systems, the miniaturization of the system was attempted to make it more portable and user friendly. A detailed analysis of the tissue oxygenation measurement is described.
The instrument design allows the collection of optical data making use of simple electronic components and an efficient circuit and the signal output is displayed as oxy-hemoglobin and de-oxyhemoglobin concentrations ([HbO2], [HHb]) in the blood. To calculate oxy and de-oxy hemoglobin concentration, Modified Beer Lambert law was used. The use of ARM7 based processor is stressed since ARM processor supports real time solutions and multi-tasking. Thus, this application could be merged with many other bio-medical applications in the future. After testing on volunteers and initial gathering of data, the experiment results imply that the equipment is not only user-friendly but also efficient and accurate.
This system based on embedded structure could be applied broadly in the determination of tissue parameters without invading the body. This system could also prove to be useful in diagnosis of breast cancers and brain tumor.
 DATA ACQUISITION-A DCS APPROACH
PROJECT GROUP: SUVRAT ALSHI, ROHAN SHELAR, PRANIL DEONE
GUIDE: PROF. M.M. PARMAR
Abstract: Data acquisition systems are very powerful and standalone systems. When these systems are coupled with a network, it gives greater amount of edibility and robustness to the data acquisition systems. This report deals with the very idea implemented using real time extended kernel and with the concept of distributed computing in mind. Here a simple car model as a work-bench and a tcp/ip protocol using sockets are implemented with xenomai (real time extension of standard linux kernel) with the priority levels set accordingly.
 VEHICLE TRACKING SYSTEM
PROJECT GROUP: SURESH BOMMISETTI SIDDHARTH DEEKSHIT, HARSHAD DESHPANDE
GUIDE: PROF.M.M. PARMAR
Abstract: Our project Vehicle Tracking System is a standalone system which helps display the co-ordinates of a vehicle on a remote PC loaded with internet. When the stand alone system is armed and loaded, the location of a vehicle can be tracked by sending vehicle coordinates through a two-way paging link or a cellular telecommunications link to a base station. The stand alone system continuously transmits the co-ordinates of the vehicle through the internet using GPRS connection on to a remote server and the latitude, longitude co-ordinates can be accessed.
Briefly, the vehicle tracking system includes a receiver transmitter subsystem, a GPS subsystem for locating a position and a control CPU. The CPU performs a log and locates function in which it logs the position of the vehicle and transmits it to a remote server. This operation is continued again and again as long as the stand alone system is supplied power.
The report covers the hardware and software details of the project. The hardware and software development platforms used during the implementation of the project. The future scope of this type of a Vehicle Tracking System is discussed along with the limitations of the project.
 3 AXIS SMART DRILL
PROJECT GROUP: MUKTESH WAGHMARE, PRASAD PATIL
GUIDE: PROF.G.T. HALDANKAR
CO-GUIDE: PROF.M.M. PARMAR
Abstract: Robotics has ushered a revolution in the field of technology. Within a few years of its advent it has found its place in all fields of manufacturing and process control. Our basic view to design robotic arm is to increase automation to decrease the manufacturing cost and to increase the productivity. We have chosen Cartesian co-ordinate robot as it is simple in construction and has the same accuracy over complete distance. The construction is selected so that it is conducive to industrial environments we have used a moving rail as X direction which resembles the conveyer belt scheme used often in industry. We have a fixed gantry which helps in high load conditions.
This project is a good example of how different branches of engineering are interfaced together to get a project done. Our objective is to successively move a Cartesian coordinate robot as per the given set of motions and to get the end effector at the desired destination through an assembly program and a GUI.
 DIGITAL AUDIO WATERMARKING SYSTEM
PROJECT GROUP: NARESH UNIYAL, SHREYA SHETE, NIKHIL DHOOT
GUIDE: PROF.N.A. BHAGAT
Abstract: Digital audio watermarking system is a technique for embedding additional data along with audio signal. Embedded data is used for copyright owner identification. A number of audio watermarking techniques are proposed. These techniques exploit different ways in order to embed a robust watermark and to maintain the original audio signal fidelity. This paper makes a tutorial in general digital watermarking principles and focus on describing digital audio watermarking techniques. These techniques are classified according to the domain where the watermark is embedded. A novel scheme that is able to merge digital watermarking and content authentication of digital audio is presented in this report. The embedding of additional data is performed in different signal domains. Watermark embedding is made by frequency hopping method in the Fourier domain, while the additional authentication data is hidden using the LSB modulation in the wavelet domain. The perceptual transparency is achieved using the frequency masking property of the HAS. The scheme obtains high robustness against standard watermark attacks and localizes accurately tampered parts of the audio clip.
 REMOTE VIDEO MONITORING OVER THE NETWORK
PROJECT GROUP: CHITRA MAHALE, NISHA PANDEY, PRIYANKA PAWAR
GUIDE: PROF.N.A. BHAGAT
Abstract: The main objective of the project is to develop a security system. The main intention of this project is to design a security system which senses a human presence in prohibited area and sends message to control room through GSM. During this transmission process of message via GSM the live videos will also be visual to the control room personnel. The system is defined for security purposes especially in the case of sudden attack. It provides wireless video surveillance of a remote location. The system comprises of a remote unit camera and a video encoder /decoder linked to a transceiver which would enable the transmission of video data over the network. The system would provide faster transmission rate with great accuracy.
 RETIMINDS ROBOT
PROJECT GROUP: ABHISHEK DESHPANDE, HITESH BECHRA, PIYUSH JAIN
GUIDE: PROF.P.V. KASAMBE
Abstract: The ability to recognize humans and their activities by vision is key for a machine to interact intelligently and effortlessly with a human-inhabited environment. Because of many potentially important applications, “looking at people” is currently one of the most active application domains in computer vision. The scope of the real time processing is limited to work on recognition of human body; it does not include work on identifying of human faces. The other two part of our project is mine detection as well as combustible gas detection.
 VARIABLE FREQUENCY DRIVE AND SMART SENSOR NETWORK FOR INTELLIGENT PROCESS CONTROL USING PLC AND SCADA
PROJECT GROUP: MAYUR AGARWAL, ASHISH SHRIVASTAV, RAMAN GAYAKWAD
GUIDE: PROF.P.V. KASAMBE
Abstract: We have created a machine vision-based real-time Quantity detection system for bottle filling station. A bottle filling station comprises of following 4 stages:
- Bottle Filling
- Bottle Liquid Level (Quantity) Detection
- Bottle Capping
- Bottle Labelling
For quantity detection system images are first captured by a camera situated on the assembly line. Bottle image is processed using OpenCv libraries. Depending on the relative count of pixel values of the processed image and after comparing with the reference image, the bottle is validated.
The system is tested on a hardware platform consisting of a TI OMAP L138 and also Atmega 8535 chip along with other supporting peripherals. The experimental results obtained so far show great promise in the near future. The system can verify 2 bottles per second for level detection.
The further part of the project can be extended to the capping station where the controller will read the signals from the DSP and reject the imperfect bottle off the capping station.
 POWER GENERATION USING SOLAR AND WIND ENERGY
PROJECT GROUP: JAGRTUEE GAWANDE, TANMAY POTNIS
GUIDE: PROF.P.V. KASAMBE
Abstract: The aim of this project is to design a system in order to utilize renewable sources of energy such as wind and solar energy for generating electricity.
The proposed system is supposed to use a solar panel and a wind turbine model to generate electric energy. This energy will be stored in a battery in DC form. As and when required, this energy can be utilized for appliances by passing through an inverter to convert into AC. It can also be used to connect to the grid for continuous power supply.
 VIRTUAL INSTRUMENTATION USING LABVIEW
PROJECT GROUP: SHWETA U. SHETTY, SHILPA SREEKRISHNAN, SHREELAXMI GOVINDAN
GUIDE: PROF.P.V. KASAMBE
Abstract: The report gives information about the implementation of a PC based virtual instrument (VI) for the experimental setup of a DC motor. Using the VI, the setup can be virtually and remotely controlled. The proposed VI is based on the LabVIEW (Laboratory Virtual Instrument Engineering Workbench) language under the workbench of Windows XP.
Reading inputs, graphically representing data and manipulating the circuitry through the software itself, all in real time, makes this application of a virtual laboratory extremely efficient and a very desirable need of our time. The real time inputs (voltage and current) are taken from the external circuit with the help of the analog channels on the NI ELVIS kit and the DAQ (Data Acquisition Card PCI-6251). These devices are used with LabVIEW to take input from the external circuit.
This can be extended to implementing various laboratory experiments virtually, along with their records, establishing a virtual laboratory and making experiments very simple to perform. It reduces large chunks of workload and hardware to make the system more efficient both with respect to time and accuracy.