several computer vision projects using computer vision, deep leaning, and other CV techbiques
Cifar10 classification: Implementing KNN classifier from scratch and chose best K value, use logistic regression as a multiclass classification using OVR algorithm from scratch, and use SVM different experiments and hyper-tuning.
Caltech Bird classification: Design and train fully connected neural network, adapt ResNet-101 network using transfer learning to train the first model by removing FC layers, but leaving the other layers intact, and visualize the filters learned by ResNest-101 network.
Plants Leaf classification: Using the first 2 blocks of VGG-16 and add extra layers to create a special version of a CNN network to classify the images according to the number of leaves in the planet images, and then reformulate the model as a regression problem, and fine tune the parameters.