Defining a network structure

The hardest part if to develop an intuition to decide on what layers you will include in a given network since the network will perform different depending on the complexity of the task, some recommended things to try are: Change the number of convolutional layers Increase the size of convolutional kernels for larger images Change…

Data augmentation

In order to improve models without adding new data, we can do data augmentation, this will work well for images that do make sense to make transformations on top of, so for example it is not a good idea to do it for text or numbers because text the orientation of characters in these cases…

Computer vision algorithms

ORB Algorithm: Oriented Fast and Rotated Brief, creates feature vectors from detected keypoints and is invariant to rotations, changes in illumination, and noise. HOG Algorithm: Histogram of Oriented Gradients works by creating histograms of the distribution of gradient orientations in an image and then normalizing them in a very special way. This special normalization is…

100DaysOfMLCode Index

  100DaysOfMLCode Index Attention mechanisms Batch size CNNs Computer Vision Conda Data Augmentation Defining a network structure Downloading Datasets Dropout Embeddings FastAI Filtered Images Facia-Keypoints-Detector notes GPU States High bias & high variance Hyperparameters Image Captioning Project Notes Intro to Pandas Lab Jobs in Computer Vision Layer Shapes Learning Rates Localization LSTM cells Momentum Machine…