The Tomato Leaf Disease Detection dataset, available on Kaggle, comprises over 20,000 labeled images across 10 classes, facilitating the classification of various tomato leaf diseases. This extensive collection supports the development and evaluation of machine learning models aimed at accurately identifying and diagnosing tomato plant ailments, thereby contributing to advancements in agricultural disease management.

This project aimed to classify videos using a Vision Transformer (ViT) model applied to the UCF50 dataset, which contained videos of 50 different action classes. The process began with extracting and preprocessing frames from videos, resizing, and normalizing them to create a consistent dataset. The extracted features and labels were saved as .npy files for future use in training. The Vision Transformer model was then designed with custom layers for tubelet embedding and positional encoding to handle the spatial-temporal information in the video frames.

The Vision Transformer (ViT) stands out as a groundbreaking architecture that redefines how we approach computer vision tasks. By leveraging the Transformer model using TensorFlow, ViT processes images as sequences of patches, offering a compelling alternative to traditional Convolutional Neural Networks (CNNs). We walked through the implementation of ViT for classifying flower images.

The goal was to combine Detectron2 for pose estimation and LSTM for action classification and we have built a powerful human action recognition system.

This code tunes the hyperparameters for an SVM regressor using the California housing dataset and evaluates the model's performance using Root Mean Squared Error (RMSE).

The objectiv was to train a Decision Tree on the moons dataset, fine-tune it using grid search, evaluate its performance, and visualize the resulting decision boundary.

The goal was to perform regression using Decision Trees, visualize the dataset and the resulting tree structure, providing a clear understanding of the model's decision-making process.

The primary goal was to train and visualize a Decision Tree classifier using the Iris dataset. Decision Trees are intuitive and powerful models for classification and regression tasks. By visualizing the Decision Tree, you can gain insights into how the model makes decisions based on the features of the dataset.

The goal was to generates a linearly separable dataset, trains a LinearSVC, SVC with a linear kernel and SGDClassifier, and plots their decision boundaries to see if they produce roughly the same model.

The goal was to implement Batch Gradient Descent with early stopping for Softmax Regression. The early stopping mechanism monitors the cross-entropy loss and stops training if the validation error does not improve for a specified number of epochs (patience).

The Titanic dataset is a classic machine learning problem. It provides information about the passengers aboard the Titanic, and the goal is to predict whether a passenger survived or not based on various features such as age, gender, class, and more. This project is an excellent introduction to data cleaning, feature engineering, and model building. We prepared the data for training and then train a RandomForestClassifier.

In this project, we have done MNIST Classification with Data Augmentation Using KNeighborsClassifier

In this project, we built a machine learning classifier for the MNIST dataset, aiming to achieve over 97% accuracy on the test set. The MNIST dataset is a classic dataset in the field of machine learning, consisting of 70,000 images of handwritten digits (0–9). Each image is 28x28 pixels, and the task is to classify each image into the corresponding digit.

In this project using TensorFlow library, we analyzed images using semantic image segmentation. Semantic segmentation's goal is to categorize each pixel in an image into a class or object. We analyzed football (or soccer) player positions on the field. There are 512 images in the set, together with JSON file containing image information.

We worked on how to train DCGAN Model using PyTorch to generate images.

Here, the goal was to perform transfer learning for image classification using the PyTorch deep learning library.

Here, we used PyTorch to detect objects in input images using seminal, state-of-the-art image classification networks, including Faster R-CNN with ResNet, Faster R-CNN with MobileNet, and RetinaNet. Also performed real-time object detection in video streams.

Here we used PyTorch to classify input images using seminal, state-of-the-art image classification networks, including VGG, Inception, DenseNet, and ResNet.