Paper Walkthrough — Segment Anything — Today’s paper walkthrough it is going to be visual! We will analyze Segment Anything, a paper by Meta’s AI research team that made headlines not only in the research community but also by all sorts of deep learning practitioners and advocates. Segment Anything introduces the task of promptable segmentation, it…

Paper Analysis—Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning — In today’s paper analysis we will have a close look into the paper behind BYOL (Bootstrap Your Own Latent). It provides an alternative to contrastive self-supervised learning techniques for representation learning removing the need for a large corpus of negative samples and gigantic batch sizes. Furthermore it is a landmark…

Paper Summary: Grounded Language-Image Pre-training — Today we will dive into a paper that builds upon the great success of CLIP in language-image pre-training and extends it to the task of object detection: GLIP — Grounded Language-Image Pre-training. We will cover the key concepts and findings of the paper and make them easy to understand by…

Paper Summary— Learning Transferable Visual Models From Natural Language Supervision — In this article we are going through the paper behind CLIP (Contrastive Language-Image Pre-Training). We will extract key concepts and break them down to make them easy to understand. Further, images and data graphs are annotated to clarify doubts. Paper: Learning Transferable Visual Models From Natural Language Supervision

An Advanced Guide for TensorFlow — I want to share with you a neat little trick on how I test my multi GPU training code on a single GPU. Motivation I guess the problem is obvious and you probably experienced it yourself. You want to train a deep learning model and you want to take advantage of…

Multi-dimensional Fourier transformations in convolutional neural networks — Convolutional neural networks (CNNs) are widely spread these days. Regardless of their success, convolutions are inefficient. The sliding window requires many computations and limits the size of the kernel. At the same time, a small kernel, typically between [3,3] to [7,7], limits the perceptive field and many layers are required…

A Resource-Saving and Performance-Competitive Alternative to Vision Transformers — This posts intents to provide a brief overview of the MLP Mixer introduced in the paper MLP-Mixer: An all-MLP Architecture for Vision by I. Tolstikhin, N. Houlsby, A. Kolesnikov, L. Beyer et al. [1]. …

A Step-by-Step Guide — You’d like to train your deep learning model on your own computer or server, you don’t want to deal with the resource limitations of free environments such as Google Colab or Kaggle and you don’t want to spend hours on installing all drivers and packages required to run your training…

Understand your Hardware to Optimize your Software — In this post I will show you the basic principles of tensor processing units (TPUs) from a hardware perspective and show you step-by-step how you can perform accelerated distributed training on a TPU using TensorFlow to train your own models. Outline Introduction 1.1. Tensor Processing Units 1.2. Distribution Strategies Implementing Distributed…