On Tuesday, we showed our client our Tumblr prototype, with a simple front-end mockup. He found it interesting and wanted to see how effective the labeling is in a video context. Since the launch of Battlefront is coming up, he also wanted us to pivot to the Battlefront franchise.
Hence, we began to look at how effective Google Vision API is on Star Wars related images.
From the above image, we can see that the labeling is not very good since it cannot identify the meaning or meaningful objects in the picture. However, the Web Entities result showed many relevant results, such as Stormtrooper being identified. However, the web entities function is reliant on Google’s search results and user-related data. Hence, it is difficult to have those when the content is new (recent uploads).
We met with a marketing analyst and received huge insights on how the player base is separated and the tools they use for data collection.
We have also finalized our branding materials! Yay!
For next week, we will focus on building the Youtube scraper and see if we can obtain good information from parsing Battlefront videos. We will also build an interactive front-end for our Quarter walkarounds so that users can interact with the prototype.