Projects
Projects
Embedded Files
Audio Intent Classification and Emotion Classifier for IVR
Audio Intent Classification and Emotion Classifier for IVR
TechStack: Python, Numpy, Librosa, sklearn, NLTK, seaborn, Ipython, Keras
Utilized the Skit-S2I dataset, which focuses on intent classification in the banking domain.
Implemented a sentiment analysis system using the CREMA-D dataset. Through extensive data visualization, augmentation, and feature extraction (MFCC, Zero Crossing Rate), a CNN model was trained and tested with varying batch sizes and epochs.
Future Enhancements: Hyper parameter tune the model parameter to get optimize and increase the model efficacy further.
Text Detection and Script Identification in natural scene videos and imagery
Text Detection and Script Identification in natural scene videos and imagery
(Research Project)
(Research Project)
TechStack: YOLOv5, Makesense.ai, Tensorflow, Machine Learning, Google Colab
Learning about the YOLO algorithm for text localization and object detection by using 3 datasets of images having English, Hindi and Kannada text.
Building a Machine Learning model for script identification in a video,
Created a web application with a user interface for identification of the script in a video streaming online.
A Relative Study on LULC Classification using ResNet-50, VGG-16 and VGG-19 models.
A Relative Study on LULC Classification using ResNet-50, VGG-16 and VGG-19 models.
(Research Project)
(Research Project)
TechStack: Machine Learning, Python, Tensor-Flow
We use the EuroSAT dataset which is highly used for land use and land cover (LULC) classification and data-preprocessing is done before training the models.
Using the EuroSAT dataset that has over 27000 images and 10 classes, we trained 3 different models ResNet-50, VGG16, and VGG19 models
These models are compared in terms of Precision, Recall, F-measure, and Accuracy.
Comparing the obtained results with ground truth and other algorithms for validation
Scribbler - A website to align visually impaired students with scribes
Scribbler - A website to align visually impaired students with scribes
TechStack: HTML, CSS3, JavaScript
A desktop website is developed using Python to align volunteered scribes with visually impaired students.
Based on location and examination details provided, they are paired up to take the exam.
Python - Django is used to build the graphical user interface.
Various web responsive pages are developed using HTML, CSS3, JavaScript, and Bootstrap.
scribe is awarded a certificate issued by the website after helping a student.
Static Web Page Development
Static Web Page Development
TechStack: HTML, CSS3
This minor project is a description page of computer hardware and how they work.
Using, HTML5, CSS3 and Bootstrap total of 6-7 html pages were created.
Navigation bars, page layout and image carousal were included after detailed planning.
To view the web pages - Click here
Page updated
Google Sites
Report abuse