Described in this article is a simple deep convolutional network (CNN), with a state-of-the-art EfficientNet backbone and an attention attention mechanism, to classify the severity of diabetic retinopathy in retinal images. The model can be trained using the publicly available diabetic retinopathy dataset on Kaggle.
In this post, the attention mechanism for neural networks is explained. The explanation has two parts. The first part deals with the intuitive understanding of the attention mechanism in a neural network. The second part implements an attention function using Tensorflow-Keras in Python3.
Described in this blog post is a prototype deep neural network model for predicting the genome sequence of the most prevalent SARS-CoV2 mutant variant.
The various mutant SARS-CoV2 variants' genomic data were sourced from the Nucleotide database of the National Library of Medicine (NLM).
The deep neural network prototyping was done using Python and Keras recurrent neural network API.
The interactive dashboard above is built using Altair, a declarative visualization library in Python. This tool visualizes the progress of India's COVID19 vaccination drive.
Here is a deep technical dive into building this visualization tool.
In this code example, the input images using the phase contrast microscopy images and their corresponding mask labels, from the Sartorius cell instance segmentation challenge data-set in Kaggle, are horizontally and vertically flipped using OpenCV.
In this code example, the input images using the phase contrast microscopy images and their corresponding mask labels, from the Sartorius cell instance segmentation challenge data-set in Kaggle, are rotated to a specific angle (in degrees) using OpenCV.
In this Python3 code example, a set of MRI images from one scanning session is resampled to match the imaging axis of a reference scan. The DICOM images are handled using PyDICOM. The resampling function is implemented using SimpleITK.
An example use case for this code example is as follows. Consider two MRI scans, A and B, with scan A imaged along the sagittal (longitudinal) plane and scan B along the axial (horizontal) plane.
If the reference image is set as scan A, then the scan B images will be transformed to appear as image slices along the sagittal plane; instead of the original axial plane used during the scan acquisition.
Similarly, if the reference image is set as scan B, then the scan A images will be transformed to appear as image slices along the axial plane; instead of the original sagittal plane used during the scan acquisition.
The example notebook is hosted in Kaggle. The MRI scans used in this notebook are from the RSNA MICCAI brain tumor classification dataset.
This blog article on Twitter saliency filter analysis introduces a few broad concepts to test machine vision tools. Described here is an end-to-end automated statistical analysis tool that is used to analyze the Twitter saliency filter. The aim is to accelerate the development of scalable, automated testing of machine vision algorithms for possible biases.
Moad Computer is an actionable insights firm. We provide enterprises with end-to-end artificial intelligence solutions. Actionable Insights blog is a quick overview of things we are most excited about.