Bookmark and Share

Technical Articles

This article analyzes the three main computation blocks of the FFT mixed-radix in a step-by-step approach, in both theory and implementation


For the foreseeable future, we will be seeing more and more video cameras built into more and more devices, with shakier use-case scenarios.

This practical case study analyzes image convolution kernels (used in many image processing tasks) written using OpenCL.

This article builds up the background for the 1D complex to complex FFT algorithm, pointing out the limits of DFT using direct computation.


An interview with the core U.C. Berkeley developers covers the history, current status and future plans for Caffe and an upcoming tutorial.

The term "sugarcube" represents an important trend in vision technology towards miniaturization.

The LPIRC focuses not only on accuracy but also the speed, cost and power consumption of the image classification solutions tested.

Memory bandwidth is the main bottleneck for implementing power-efficient algorithms for computer vision.

By far the most performance-intense functionality in the neural network are the convolutions themselves.

Computer vision and computational photography are intrinsic aspects of the Internet of Things (IoT).