MULTI-DRONE MISSION FLANNING AND VIDEO STREAMING
An Unmanned Aerial Vehicle (UAV), commonly known as a drone, is an aircraft without a human pilot onboard. UAVs are a component of an unmanned aircraft system, which includes a UAV, ground –based controller and a system of communication. The project is to build a prototype drone and achieve navigation in the Non-GPS zones using the feature matching technique with help of a live video, live video frames are matched with the feature matching technique with a preloaded set of images. The live video streaming was done using Gstreamer and feature matching technique. In this project challenges in building the quadcopter and connecting multiple drones from a single ground station using the mission planner and video streaming with on board recoding are addressed. Here the live streaming is achieved using GStreamer application and the video recorded is on the Rasberry Pi on board storage for image matching from computing the way points for the auto flight operations.
- Akhil Sai Rudra(1602-15-737-004)
Our application is to create a website through which we can temporarily hire a travel guide. Generally there are some tourist pilgrimage places, which needs local guide who can direct us, throughout their visit. In this application tourist can book the guide in advance. First Guide has to be registered in the website while registering he/she has to enter the some basic information like Name, Address, Languages he/she known, Contact Number, Email-id, Password after entering all details we will send an OTP for security purpose.
In this developed website best tourist places are listed for ease of tourists. This web application consists of different modules such as registration of tourists, login module and listing of the guides based on tourists requirements from which they can opt one guide.
- RISHIKA REDDY N, 1602-16-737-093
- K SAI SUJITH, 1602-16-737-098
- NALLAPU SRIKAR, 1602-16-737-109
The objective of this application is to create a SMART BULB which can be controlled from anywhere in the globe using Internet Of Things. With the help of this, we can control the device even when we are away on vacation with ease. We use the Arduino Software (IDE) which will allow us to write programs and upload them to NodeMCUboard. We use BLYNK on the mobile end to leverage the connectivity between the user and the bulb. This app is a platform with IOS and Android apps to control Arduino, Raspberry pi and Node-MCU.
- AGARALA.HARSHITH, (1602-16-737-015)
- S.A S IRSHAD, (1602-16-737-017)
INTELLIGENT SAFETY SYSTEM
This application ensures the safety of people in emergency. This helps to identify and call on resources to help the one out of dangerous situations. This reduces risk and brings assistance when we need it and help us to identify the location of those in danger.
The best way to minimize your chances of becoming a victim to accidents, robber etc. is to identify and call on resources to you out of dangerous situations. Whether you are in immediate trouble or get separated from friends during night out and don't know how to get back to home, having such an app on your phone can reduce risk.
This app contains features like sending a default message, making a phone call and sharing live location.
- Y.Apuroop - 1602-16-737-008
- Meharaj - 1602-16-737-024
- P.Naresh - 1602-16-737-025
SAFE THAN SORRY
- Always know the narrow and dingy paths you might be led into.
- Know how to get back onto a relatively crowded area, in case open might need help.
- Get to the nearest help i.e. Police Station, Hospital or close by establishments like malls, hotels, pharmacies etc.
STS - Safe Than Sorry is an app to prevent users from travelling through isolated or unfrequented routes. According to Times Now news, records indicate that a large number of rapes, molestations and murders occur in areas which are inaccessible by Police Control Room vans. Hence, the primary motivation, is to redirect the user onto a more reliable locations.
STS calculates the safety of a route based on the commercial establishments present along the route, taking into account their timings, the user’s estimated time of arrival to the destination and safety quotient (probability of finding help) of the type of establishment.
- KiranmayiAnupindi (1602-16-737-078)
- MuthyalaLahari Reddy (1602-16-737-020)
- Viswanatham Sarika (1602-16-737-102)
EVENT MANAGEMENT APPLICATION
This application intend to solve the problems of propagating news and information, and also alleviate the problem of traditional event managing procedures such as lots of paper work, or long queue at the registration desk. The objective of this project is to develop an android application which provides interesting news and events. Moreover, users will be able to manage their event participation, such as reserving their seats in events, registering at the event site, and so on. Modules of the application are User Authentication, Real Time Event Data Flow, Notifications System for the Events, Classified Events, Event Attendees and Bookmark, Event Member Management, Event Coordinator Management, Event Analytics(Extension).
- Android Studio - The Development Platform used for Android Application
- Firebase - Firebase is Google’s mobile platform that helps you quickly develop high-quality apps
- Maven - Dependency Management
- SAICHARAN.K (1602-16-737-095)
- SANTOSH KUMAR(1602-16-737-101)
- PRAVALLIKA.K (1602-16-737-091)
CRIME RATE ANALYSIS
Project is on new pipeline for improving querying and analysis on large data sets ( i.e for crime rate analysis).Though average crime rates in India have been in the decline for the last few years (www.academike.com/crime-crime rates). It is still useful to many groups, such as law enforcement, city officials, home buyers, etc., to be able to predict where and when crime will occur. We develop a model to predict future crime incidence at a future time given a geographical location, leveraging historical crime data from the cities of Hyderabad, Delhi etc.. We take a Bayesian non-parametric approach to the problem of interpolation on crime data from these cities. We explore Gaussian Processes with simple closed form kernels. We compare these models to current baseline approaches which consist of linear regression on crime data, partitioned by region. By using spark and hive which is already built on top of Hadoop, we can easily create a new pipeline for doing analysis on large data sets i.e. in this case to analyse or go over the statistics of previous crime history records. By developing this model, it will be easy to be able to not only predict the crime but also to safeguard the people in advance. To develop this model, it will be using the apache Hadoop, hive and spark with a deep learning model using TensorFlow. As a problem driven project a significant portion of this project will include implementing various regression algorithms.
AREA: BIG DATA ANALYSIS- USING SPARK,HADOOP WITH PIG, HIVE FRAMEWORK
- T.K.Aishwarya– (1602-16-737-001)
Vehicle Detection and Counting in High-Resolution Aerial Images Using Convolutional Regression Neural Network
Vehicle detection and counting is important for many applications such as surveillance, traffic management, and rescue tasks. The ability of on-line monitoring of vehicles distribution in the urban environments prevents traffic jams and congestions which in turn reduces air and noise pollution. However, this task is a challenging one due to the small size of the vehicles, their different types and orientations, and similarity in their visual appearance, and some other objects, such as air conditioning units on buildings, trash bins, and road marks. These methods are either based on shallow learning or deep learning approaches.
Vehicle detection and counting is a challenging task due to many reasons such as: small size of the vehicles, different types and orientations, similarity in visual appearance of vehicles and some other objects (e.g., air conditioning units on the buildings, trash bins, and road marks), and detection time in very high resolution images is another challenge that researchers need to take in consideration. The number of the cars detected has been determined by the estimation of the detected regions. Hyper feature map that combines hierarchical feature maps have been used in an accurate vehicle proposal network (AVPN). Vehicle location and attributes have been extracted by the proposed coupled regional convolutional network method which merges an AVPN and a vehicle attribute learning network. Fast and Faster R-CNN have been explored .In order to overcome the limitations in Fast and Faster R-CNN, a new architecture has been proposed. They have improved the detection accuracy of the small-sized objects by using the resolution of the output of the last convolutional layer and adapting anchor boxes of RPN as feature map.
The approach we follow to solve vehicle detection and counting problem by a fully convolutional regression network (FCRN). During training, an input image and its corresponding ground truth are given to the FCRN where the goal is to minimize the error between the ground truth and predicted output. During inference, the output of the trained model goes under an empirical thresholding after which a simple connected component algorithm is used for returning the count and the location of the detected vehicles.
- B.Sireesha - 1602-16-737-317
- K.Ankitha - 1602-16-737-319
PLANT LEAVES DISEASE IDENTIFICATION WITH CONVOLUTIONAL NEURAL NETWOK.
The Fungal diseases that are very common in plant leaves is one of the main causes for drop in the quality and the quantity of the agriculture production. They affect the quality of the plants and their products. They not only influence the economic importance of the plants and its products but also abate their ecological prominence. Main aim of the Project is to develop an appropriate and effective method for identifying the exact disease which can help in finding a suitable solution. Over the last few years, due to their higher performance capability in terms of computation and accuracy, computer vision, and deep learning methodologies have gained popularity in assorted fungal diseases classification. Therefore, a convolutional neural network (CNN) is proposed for the classification of infected leaves. The results from this model achieve better classification accuracy when compared to existing approaches.
By this project, we identify the type of diseases in plant leaves, which will help in taking necessary action to reduce or remove the diseases in leaves, thereby eventually resulting in better plant productivity. By controlling the biotic factors which cause severe losses in the crop yield, we can enhance the productivity and quality of the plants and its products. Use of technology in addressing plant diseases give better results with far more accuracy. Computer vision with machine learning methodologies has outperformed in solving a number of plant leaves disease problems including pattern recognition, classification, object extraction etc. Therefore in this work, we propose an innovative model named as CNN for the classification of leaves infected from fungal diseases. The presented model is also computationally efficient and simple. We used tomato and corn leaves in this model to predict the diseases in those leaves. Hence the images of these leaves are taken as raw input and by performing convolution operation we then applying those weights to rectified linear unit(ReLu) activation function. After that to reduce size of image which improves memory and model efficiency we were used max pooling in pooling layer and then by using fully connected layer the output values are obtained in an one dimensional array, Later softmax function is used to get the probability of every class for the images we have trained and tested. However, it may be noted that this model has a limitation as it has problem of over-fitting and its mostly time consuming because it has to take a large dataset for training.