Image Captioning is the process of generating textual description based on the objects and actions of an Image. It uses both Natural Language Processing and Computer Vision to generate the captions. Generally Image Captioning is dividing into two modules. 1st is image based model which extracts the features and noise out of Image. And 2nd is language based model which translates the features and objects given by our image based model into natural sentence.
For Image based model - We used Convolutional Neural Network (CNN) and for language based model - We used Recurrent Neural Network (RNN). For CNN we used 3 different Algorithms like VGG16, ResNet50 and ResNet101. And for RNN we used 2 different Algorithms like LSTM and Glove Embedding and Stacked LSTM Layers.
By Using thses algorithms we made 3 models - 1st is VGG16 for features Extraction and 1 LSTM for Sequence model. 2nd is ResNet50 for features Extraction and 1 LSTM for Sequence model. 3rd is ResNet101 for features Extraction and Glove Embedding and Stacked LSTM Layers for Sequence model. Out of these you can use any of the model to generate captions. We used popular python web framework Flask for deployment our model and use it as a Web application. At the we concluded that the 2nd model give more accuracy and the captions generated by this model are also very good.
Institutes's Information Interchange & Communication Application - ( IIICA ). This is an android application. This application is android based application which gives the platform between the Institute and students to share Queries, share knowledge and also used for consulting.
The Technologies we used were Android studio, Java and for Backend we used PHP. We made this application in Local Host. So, we used Xamp Server. For transferring the user data to the server we used PHP Script.
For fetching the data from the server which is in the JSON format, we convert it into a javascript objects. We parse the JavaScript object to display the data into Application.
Automatic Plant Irrigation System - ( APIS ). This was an arduino based project. In this project we used the arduino with Wifi sheild and Soil Moisture Sensor and Humidity Sensor to detect the Moisture and Humidity of the Soil.
Whenever the moisture of the soil increase of it's set limit the notigication will pop in the device from which the hotspot is connected to the Arduino Wifi sheild. We can turn off and On the pump Automatically as well as manually by changing the setting in the code. We used MYMQTT app to as a interface to operate the Module.