keywords: projections, image analysis, clustering, principal component analysis, metrics, classifying spaces
image source: https://www.amazon.com/Latvie%C5%A1a-Cimdi-Maruta-Grasmane/dp/9934141418
The Latvian tradition of knitting mittens is well-established, up to the point of thorough books with descriptive and precise directions / images for how to knit certain patterns. Mittens and their properties are often associated with geographical regions of Latvia, and the goal of this project would be to relate mitten patterns with regions of Latvia by using a clustering method, such as principal component analysis, when interpreting a mitten as a vector in appropriate 'mitten space'. Key parts of this project would be to first uniformly interpret mittens as vectors, then to define what 'pattern' means for a mitten. In the classical sense a pattern covers color, size, thread properties, self-similarity, rotation, etc. A starting point for the data set is 'Latvieša cimdi' by Maruta Grasmane, which has a broad collection of mittens, associated to regions, and reproduced in a digital grid, which have been shared as digital files.
keywords: image analysis, motion tracking, urban planning
image source: http://nationaltrafficsurveys.com.au/manual-counts/
Given a road intersection, there are a limited number of possible "paths" a car, bike, person, etc. can take when passing through it. For example, at a 4-way intersection, a car can enter from the north and continue straight south or turn east at the intersection. Surveying key intersections in Riga and determining popular transportation patterns is important for the Riga city council to predict / alleviate congestion and to plan transport network expansions. This goal of this project would be to develop a tool to which a video is input and paths + frequency + type of transport is output. The video may be assumed to be stable (with fixed markers).
keywords: Raspberry Pi, image analysis
image source: https://www.arducam.com/raspberry-pi-camera-pinout/
This is an open-ended project, with a key component being the Raspberry Pi (which will be provided). This project involves learning to use it, setting it up in different real world settings, and if applicable, reading images from the associated camera module. One example is setting it up with a view of a city street and measuring the speed of passing cars, or how many cars cross at a red light. Another example is comparing the number of people who cross at a red light to the number of people present, or the width of the street, or the time of day. A third example is using the Raspberry Pi on a moving object (like a bicycle) to produce real time feedback, about things such as width of the road, warning signs, inteference, etc. All hardware is provided.
keywords: network theory, graphs, urban infrastructure, traffic analysis
image source: https://www.trafficplans.ie/what-is-a-traffic-management-plan/
The layout of streets, for cars / bikes / pedestrians, is created together with expected routes that participants of traffic take. In particular, green lights are often synchronized to ensure cars / bikes do not have to brake and can quickly get from one point to another. This project would examine existing traffic corridors in Riga and would suggest how to imporve them, and where to create new ones. Part of the project could be data collection of traffic participants at key locations, either by hand or by analyzing video frames.