Fahad Hassan's Projects
Code to identify Colors in an image using color dataset. It compares color of the particular area of the image with all the colors in the dataset and gives accurate color prediction.
I've tried to find out the weak areas where we can work to make more profit. What all business problems we can derive by exploring the data. Technologies used are Python, Numpy, Pandas, Matplotlib, Seaborn.
Performed Exploratory Data Analysis on Terrorism Dataset to find useful information related to world wide terrorist attacks. Tools used are Python, Numpy, Pandas, Matplotlib, Seaborn and Plotly.
Task#4 as an "IoT and Computer Vision" intern at THE SPARKS FOUNDATION. A real time Face Mask Detector using MobileNet, OpenCV, Tensorflow and keras, sklearn, Numpy, and Matplotlib. In this project, I've used transfer learning using 'imagenet' weights and further trained my model on DNN. I've implemented YOLO to make it a real time face mask detector using webcam.
Config files for my GitHub profile.
Task#1 as an "IoT & Computer Vision" intern at THE SPARKS FOUNDATION. An Optical Character Recognition (OCR) which recognize text in images using Keras_ocr and plot images and text using Matplotlib.
A supervised Machine Learning model that predict student score percentage on the basis of study hours. The model uses Linear Regression to predict percentage score of a student. The linear regression is provided by Scikit-learn.
In this model, I've achieved training accuracy of 90%, while validation accuracy of 95%. The model is trained on the training dataset of 2520 images and validation dataset of 372 images. The model used DNN and CNN layers with Dropout to trade off the variance and bias.
The model uses 50K movie reviews labelled as 'positive' and 'negative' that implements a Sentiment Analysis model using LSTM and GloVe(Global Vectors) Word Embeddings. There is 100-dimmensional GloVe(Global Vectors) word Embeddings is used. It is an unsupervised learning algorithm used for the similar representation of words having same meanings. LSTM(Long Short-term Memory) is a type of RNN model used for handling long sequence dependencies for classification task.
Laurence Moroney's poem dataset, achieved 71% of accuracy.
Implemented IMDB_reviews dataset to to train the model for classifying positive as well as negative reviews.
Unsupervised Machine Learning model on Iris dataset using KMeans algorithm. Aim is to find opimum number of clusters for the dataset.