Articles by tag: vision

Articles by tag: vision

    Rewriting CNN

    Rewriting CNN By Arjun and Abhi

    Task: Begin rewriting the Convolutional Neural Network using Java and DL4J

    While we were using Python and TensorFlow to train our convolutional neural network, we decided to attempt writing this in Java, as the code for our robot is entirely in Java, and before we can use our neural network, it must be written in Java.

    We also decided to try using DL4J, a competing library to TensorFlow, to write our neural network,...

    Read More

    Developing a CNN

    Developing a CNN By Arjun and Abhi

    Task: Begin developing a Convolutional Neural Network using TensorFlow and Python

    Now that we have gathered and labeled our training data, we began writing our Convolutional Neural Network. Since Abhi had used Python and TensorFlow to write a neural network in the past during his visit to MIT over the summer, we decided to do the same now.

    After running our model, however, we noticed that it was not very accurate. Though...

    Read More

    Labelling Minerals - CNN

    Labelling Minerals - CNN By Arjun and Abhi

    Task: Label training images to train a Neural Network

    Now that we have software to make labeling the training data easier, we have to actually use it to label the training images. Abhi and I split up our training data into two halves, and we each labeled one half. Then, when we had completed the labeling, we recombined the images. The images we labeled are publicly available at Read More

    CNN Training Program

    CNN Training Program By Arjun and Abhi

    Task: Designing a program to label training data for our Convolutional Neural Network

    In order to use the captured training data, we need to label it by identifying the location of the gold mineral in it. We also need to normalize it by resizing the training images to a constant size, (320x240 pixels). While we could do this by hand, it would be a pain to do so. We would...

    Read More

    CNN Training

    CNN Training By Arjun and Abhi

    Task: Capture training data for a Convolutional Neural Network

    In order to train a Convolutional Neural Network, we need a whole bunch of training images. So we got out into the field, and took 125 photos of the sampling setup in different positions and angles. Our next step is to label the gold minerals in all of these photos, so that we can train a Convolutional Neural Network to label the...

    Read More

    Vision Discussion

    Vision Discussion By Arjun and Abhi

    Task: Consider potential vision approaches for sampling

    Part of this year’s game requires us to be able to detect the location of minerals on the field. The main use for this is in sampling. During autonomous, we need to move only the gold mineral, without touching the silver minerals in order to earn points for sampling. There are a few ways we could be able to detect the location of the gold mineral.

    ...

    Read More