Object Detection
Implementing the YOLO (You Only Look Once) architecture for real-time object detection in images and videos.
Image Colorization
Colorizing grayscale images with Convolutional Neural Networks.
Watermark GAN
Training an image reconstruction GAN to remove watermarks from pictures.
Human Segmentation with DeepLabV3+
Implementing the DeepLabV3+ architecture to segmentate humans from background in pictures.
Satellite Imagery Classification
Identifying deflorestation in the Amazon Basin using satellite imagery and Convolutional Neural Networks.
Instance Segmentation
Leveraging transfer learning from ResNet50 to train a Mask RCNN neural network for instance segmentation of pedestrians.
Variational Autoencoder
Generating images of numbers with a variational autoencoder and the MNIST dataset.
Satellite to Street Map
Converting satellite imagery to street maps with a Pix2Pix GAN.
Multimodal Emotion Recognition
Comparing different multimodal fusion techniques to optimize an emotion recognition system.
Image Captioning
Employing transfer learning (Xception and GloVe) to automatically extract image and text features used to train a neural network capable of captioning images.