Computer Vision Constellation Finder
Raymond Zhou | Rimika Banerjee | Ryan Lee | CS160 Group 22
Motivation
We want to provide a stargazing and constellation identification app geared towards astronomers who want to use the data for research and professional purposes. Most existing stargazing apps are intended for casual use by hobbyists and therefore do not provide anything that would be useful in a strictly scientific and professional context, but our target users need detailed scientific information about constellations and stars to conduct their research. Our app aims to bridge this gap of knowledge. Furthermore, a huge problem in astronomy research is that the equipment is heavy, cumbersome to use, and expensive. While other stargazing apps do not have the functionality of search, our app solves this problem by allowing astronomers to locate constellations conveniently with no equipment.
Our approach accomplishes three main tasks:
1) Users can view information about weather, visibility, and light pollution.
2) Users can snap a photo of the sky in order to see overlaid graphics of constellations and star charts and learn the latest academic information on objects in view. Our app will reference an external database to automatically identify it, and the information will include general properties of the star (color, size, location, heat, etc), mythological origins, star type/system classification, surrounding celestial bodies, and miscellaneous related scientific research.
3) Users can position their body according to the app's instructions to face a star constellation or star upon searching for it.
Process & Insights
Initially, we started out with a series of rough sketches: the user takes a picture of a constellation and the app identifies it and all the stars within it; clicking on a star would bring up an info page about the star. We then went and interviewed 3 students inside Berkeley's astronomy hall, ranging from freshmen to experienced graduates and hobbyists to researchers. We inquired about their stargazing habits and experiences.
After compiling the results, we realized a big issue the students had with their current stargazing methods was that they either could not afford a telescope, or they considered it cumbersome to disassemble, move, and reassemble constantly. A lesser problem was that the PhD students and other astronomy researchers we interviewed were concerned with the accessibility of scientific high-level data in this particular field, and that some casual users needed to consult textbooks to know which constellation they were looking at.
We realized then that our app was the perfect solution to the telescope issue, since of course with the app the users would no longer need a telescope or an external reference material to identify constellations. We then decided to shift our app's purpose slightly to focus on the researchers and scientific professionals, after our research showed that current stargazing apps targeted casual users, as a way to distinguish ourselves.
We constructed a low-fidelity prototype implementing constellation lookup with clickable labels for each star, star identification, weather/light pollution data, and star info lookup. After receiving feedback from the instructors and our peers' heuristic evaluations, we added more concise help screens for when the user is panning the camera to a constellation (instructions like "turn right", "move upwards", and "stop" when the constellation was in view). It was also pointed out to us that our "home" button while in the process of moving the camera was a camera icon, making the meaning of the button vague, so we changed it to a home icon. We also reduced clutter on the home screen and star info page to make them more aesthetically appealing.
For our final iteration, after another round of critique we added the "simulated night sky" feature so that users could make some limited use of this app during the daytime. We also made sure that the "home" button was in the same place for each screen to reduce user confusion and change the confirmation message when a user wants to return "home" to be less ambiguous. Finally, we made better text labels for each star to indicate more clearly that the labels could be clicked on to bring up information about the star.
Initial sketches of the app's usage and the user workflow
Detailed storyboard with 3 different potential user tasks
Unpolished low-fidelity prototype, showing a user who searches for a constellation, moves the camera to capture it, and the app automatically identifying the stars in the constellation
Final Version
Finished product. Added support for a daytime "simulated night sky" option and more detailed instructions on moving the camera to find a constellation. Shown is the user flow to take a picture of a constellation. Follow the instructions on the screen to align the camera with the constellation, and press "Take Photo" when the constellation is in perfect view. The Home button can be pressed at any time to go back to the home screen.
Shown is the workflow to view a star's data. Tap on each star's text label to bring up a zoomed-in view of the star. Tapping on the toolbars at the bottom of each screen brings up an information article about the star/constellation.
