DesignSpark Electrical Logolinkedin
Menu Search
Ask a Question

How to build your own dataset with a Neural Network Console? (Part 1)

Deep learning has become more vitally important in our daily life. The way that we unlock our phone and the intelligent LEDs on the streets are both demonstrating the example of implementing AI in terms of image recognition. There are many open sources on the internet that are about AI, including handwritten digit recognition, but how could we build our own dataset? In this article, I am sharing an application with you all, the Neural Network Console.

This article will be divided into 4 parts, the first preparation of the datasets, the second to build the Network model, the third to create the dataset and lastly, how to evaluate the model.

SONY Neural Network Console

Image from https://dl.sony.com/app/

Neural Network Console, a tool to efficiently design the neural networks in a refined user interface. And Neural Network Libraries, an open-sourced library with deep learning technologies developed at Sony. For more details, click here.

Data Set preparation

Clip a video in square format 1:1. example

A useful tool VLC player.

1. Press CTRL + P

2. Select ALL at the bottom left corner

3. Scroll down and seek the Video > Filters

4. Expand the Filters and scroll down and seek the Scene filter

5. Select the picture format as png, both Image width and height as 28, create an empty file and put the file name as the Filename Prefix and copy the path that you have created for the file and paste to the Directory Path Prefix, then choose the ratio as to how many frames do  you want to pick. Putting 1 means all frames.

Create the file and divide into groups.

Put the data set in different groups as shown in the video below.

Future development

Could a neural network console recognise a face? Yes! It would be interesting using this app as a starter.

Input different data sets and various results will be generated.

PART 2

PART 3

PART 4

Brian0925 has not written a bio yet…

8 May 2019, 8:40