artistic style – Science4Performance

Creating artistic images from Strava rides

When you upload a ride, Strava draws a map using the longitude and latitude coordinates recorded by your GPS device. This article explores ways in which these numbers, along with other metrics, can be used to create interesting images that might have some artistic merit.

The idea was motivated by the huge advances made in the field of Deep Learning, particularly applications for image recognition. However, since datasets come in all shapes and forms, researchers have explored ways of converting different types of data into images. In a paper published in 2015, the authors achieved success in identifying standard time series by converting them into images.

GPS bike computers typically record snapshots of information every second. What kind of images could these time series generate? It turns out that there are several ways to convert a time series into an image.

Spectrogram

Creating a spectrogram is a standard approach from signal processing that is particularly useful for analysing acoustic files. The spectrogram is a heat map that shows how the underlying frequencies contributing to the signal change over time. Technically, it is derived by calculating the discrete Fourier transform of a window that slides across the time series. I applied this to my regular Saturday morning club ride of four laps around Richmond Park. The image changes a bit once the ride gets going after about 1200 seconds (20 minutes), but, frankly, the result was not particularly illuminating. There is no obvious reason to consider cycling power data as a superposition of frequencies.

spectrogram

Ah! Now we are getting somewhere

The authors of the referenced paper took a different approach to produce things called Gramian Angular Summation Field (GASF), Gramian Angular Difference Field (GADF), and Markov Transition Field (MTF). Read the paper if want to know the details. I created these and something call a Recurrence Plot. All of these methods generate a matrix, by combining every element in the time series with every other element. The underling observations occurring at times $t_{1}$ and $t_{2}$ determine the colour of the pixel at position ( $t_{1}$ , $t_{2}$ ). Images are symmetric along the lower-left to upper-right diagonal, apart from GADF, which is antisymmetric.

Let’s see how do they look for on four laps of Richmond Park. We have six time series, with corresponding sets of images below. The segmentation of the images is due to periodicity of the data. This is particularly clear in the geographic data (longitude, latitude and altitude). The higher intensity of the main part of the ride is most obvious in the heart rate data. The MTF plots are quite interesting. Scroll down through the images to the next section

data1 — Raw time series of power, heart rate, cadence, longitude, latitude and altitude

From cycle ride to art

It is one thing to create an image of each item, but how can we combine these to summarise a ride in a single image. I considered two methods of combining time series into a single image: a) create a new image where the vertical and horizontal axes represent different series and b) create a new image by simply adding the corresponding values from two underlying images.

One problem is that some cyclists don’t have gadgets like heart rate monitors and power meters, so I initially restricted myself to just the longitude, latitude and altitude data. Nevertheless, as noted in an earlier blog, it is possible to work out speed, because the time interval is one second between each reading. Furthermore, one can estimate power, from the speed and changes in elevation.

Another problem is that rides differ in length. For this I split the ride into, say, 128 intervals and took the last observation in each interval. So for a 3 hour ride, I’d be sampling about once every 84 seconds.

The chart at the top of this blog was created by first normalising each series to a standard range (-1, +1). Method a) was used to create two images: longitude was added to latitude and altitude was multiplied by speed. These were added using method b). Using these measures will produce pretty much the same chart each time the ride is done. In contrast, an image that is totally unique to the ride can be produced using data relating to the individual rider. The image below uses the same recipe to combine speed, heart rate, power and cadence. If this had been a particularly special ride, the image would be a nice personal memento.

lastimage — A different take on four laps of Richmond Park

For anyone interested in the underlying code, I have posted a Jupyter notebook here.

References

Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks, Wang Z Oates T, https://www.aaai.org/ocs/index.php/WS/AAAIW15/paper/viewFile/10179/10251

Cycling Through Artistic Styles

My earlier post on cycling art provided an engaging way to consider the creative potentials of deep learning. I have found myself frequently gravitating back to the idea, using the latest code available over at fast.ai. The method uses a neural network to combine the content of a photograph with the style of an artist, but I have found that it takes a few trials to find the right combination of content versus style. This led to the idea of generating a range of images and then running them together as a movie that gradually shifts between the base image to a raw interpretation of the artist’s style.

Artistic styles

Using a range of artistic styles from impressionist to abstract, the weights that produced the most interesting images varied according to the photograph and artistic style.

My selected best images are shown below, next to snippets of the corresponding artworks. It turned out that the impressionist artists (Monet, Van Gogh, Cézanne and Braque) maintained the content of the image, in spite of being more heavily weighted to artistic style. In contrast, the more monochromatic styles (O’Keeffe, Polygons, Abstract as well as Dali) needed to be more strongly weighted towards content, in order to preserve the cyclist in the image. The selections for Picasso and Pollock were evenly balanced.

Monet from Water Lilies

Water lily mountains

Van Gogh from Starry Night

Swirly landscape

Cézanne from Sainte Victoire

Nice match with painting

Picasso from Mediterranean Landscape

Mountains become houses with doors and windows

Braque from Landscape near Antwerp

Style captured well

Dali from Metamorphosis

Fingers and eyes looking on

Pollock from Convergence

Worked well

O’Keeffe from Red Canna

Out of the fire

Polygons

Landscape still perceptible

Abstract

Smoked glass

Every image is unique and sometimes some real surprises pop up. For example, using Picasso’s style, the mountains are interpreted as rooftops, complete with windows and doors. Strange eyes peer out the background of finger-shapes in the Dali image and the mountains have become Monet’s water lilies. The Pollock image came out very nicely.

Deep learning

The approach was based on the method described in the paper referenced below. Running the code on a cloud-based GPU, it took about 30 seconds for a neural network to learn to generate in image with the desired characteristics. The learning process was achieved by minimising a loss function, using gradient descent. The clever part lay in defining an appropriate loss function. In this instance, the sample image was passed through a separate pre-trained neural network (VGG16), where the activations, at various layers in the network, were compared to those generated by the photograph and the artwork. The loss function combined the difference in photographic content with the difference in artistic style, where the critical parameter was the content weighting factor.

I decided to vary the content weighting factor logarithmically between around 0.1 and 100, to obtain a full range of content to style combinations. A movie was be produced simply by packing together the images one after the other.

References

A Neural Algorithm of Artistic Style, Leon A. Gatys, Alexander S. Ecker, Matthias Bethge

Deep Learning – Cycling Art

I’ve always be fascinated by the field of artificial intelligence, but it is only recently that significant and rapid advances have been made, particularly in the area of deep learning, where artificial neural networks are able to learn complex relationships. Back in the early 1990s, I experimented with forecasting share prices using neural networks. Performance was not much better than the linear models we were using at the time, so we never managed money this way, though I did publish a paper on the topic.

I am currently following an amazing course offered by fast.ai that explains how to programme and implement state of the art techniques in deep learning. Image recognition is one of the most interesting applications. Convolutional neural networks are able to recognise the content and style of images. It is possible to explore what the network has “learnt” by examining the content of the intermediate layers, between the input and the output.

Over the last week I have been playing around with some Python code, provided for the course, that uses a package called keras to build and run networks on a GPU using Google’s TensorFlow infrastructure. Starting with a modified version of the publicly available network called VGG16, which has been trained to recognise images, the idea is to combine the content a photograph with the style of an artist.

An image is presented to the network as an array of pixel values. These are passed through successive layers, where a series of transformations is performed. These allow the network to recognise increasingly complex features of the original image. The content of the image is captured by refining an initially random set of pixels, until it generates similar higher level features.

The style of an artist is represented in a slightly different way. This time an initially random set of pixels is modified until it matches the overall mixture of colours and textures, in the absence of positional information.

Finally, a new image is created, again initially from random, but this time matching both the content of the photograph and the style of the artist. The whole process takes about half an hour on my MacBook Pro, though I also have access to a high-spec GPU on Amazon Web Services to run things faster.

Here are some examples of a cyclist in the styles of Cézanne, Braque, Monet and Dali. The Cézanne image worked pretty well. I scaled up the content versus style for Braque. The Monet picture confuses the sky and trees. And the Dali result is just weird.