Embedded Files
Dubai Satellite Imagery Semantic Segmentation Using Deep Learning
Dubai Satellite Imagery Semantic Segmentation Using Deep Learning
Tech Stack: Python, NumPy, Pandas, Matplotlib, Seaborn, OpenCV, Albumentations, Keras, TensorFlow, Keract
In this project, I have performed semantic segmentation on Dubai's Satellite Imagery Dataset by using transfer learning on an InceptionResNetV2 encoder based UNet CNN model. In order to artificially increase the amount of data and avoid overfitting, I preferred using data augmentation on the training set. The model achieved ~81% dice coefficient and ~86% accuracy on the validation set.
Multiclass Semantic Segmentation of Aerial Drone Images Using Deep Learning
Multiclass Semantic Segmentation of Aerial Drone Images Using Deep Learning
Tech Stack: Python, NumPy, Pandas, Matplotlib, Seaborn, OpenCV, Albumentations, Keras, TensorFlow, Keract
In this project, I have performed semantic segmentation on Semantic Drone Dataset by using transfer learning on a VGG16 backbone based UNet CNN model. Artificial data augmentation techniques had been used for augmenting the dataset. The model performed well, and achieved ~87% dice coefficient on the validation set.
Which is More Fun: Weekdays or Weekends? A Spatiotemporal Analysis of GPS Data
Which is More Fun: Weekdays or Weekends? A Spatiotemporal Analysis of GPS Data
Tech Stack: QGIS, Python, NumPy, Pandas, Plotly
This project aims to analyze the spatiotemporal mobility patterns on a college campus through smartphone-based GPS monitoring over a week to identify and assess a single student's key daily actions, priorities, and unusual behaviors and compare mobility patterns on weekdays and weekends. The spatial mobility patterns are analyzed using simple mobility indicators such as speed, time spent, and mode of transportation, as well as identifying regions of concentration of specific trajectories within the study area.
Apple Scab Detection Using Deep Learning Based CNN
Apple Scab Detection Using Deep Learning Based CNN
Tech Stack: Python, NumPy, Pandas, Matplotlib, Seaborn, OpenCV, Albumentations, Keras, TensorFlow, Scikit-Learn, Pillow, Keract, Visualkeras
This project was a part of the DSE-309 Advanced Programming in Python course. The main goal of the project was to detect apple scab disease using deep CNN. The VGG16 feature extractor (pre-trained on ImageNet) with a custom classifier was trained on the augmented AppleScabFDs dataset. Considering the class-imbalance present in the dataset, the model performed well by achieving an overall accuracy of 85.71% on the test set.
Retina Damage Detection from Optical Coherence Tomography (OCT) Images Using Deep Learning
Retina Damage Detection from Optical Coherence Tomography (OCT) Images Using Deep Learning
Tech Stack: Python, NumPy, Matplotlib, Seaborn, Keras, TensorFlow, Scikit-Learn, Scikit-Image, Keract
In this project, I performed multi-class classification on the Mendeley Labeled Optical Coherence Tomography Dataset. I developed a transfer learning based VGG16 CNN model with a custom classifier and classified medical OCT scans into 4 classes, viz. NORMAL, CNV, DME and DRUSEN. The model achieved ~99% accuracy on the test set.
Animalia Kingdom Representation using Owlready2 based Ontology
Animalia Kingdom Representation using Owlready2 based Ontology
Tech Stack: Python, Owlready2
In this work, I address the task of “Animalia Kingdom Representation using Owlready2 based Ontology” for the DSE309 Advanced Programming in Python course. The Animalia Kingdom is a large group that consists of eukaryotic, multicellular organisms that are heterotrophic in nature. It is broadly divided into two classes (or groups), i.e., invertebrates and vertebrates. These two classes are further divided into multiple sub-classes. I have tried to represent the hierarchical structure of the Animalia Kingdom into an ontology using the Owlready2 package. The ontology is then visualized using the Protégé tool developed by the Stanford Center for Biomedical Informatics Research.
Page updated
Google Sites
Report abuse