Poems About Things

Generating poetry from the everyday objects around us


Poems about things is a project that generates poetry from everyday objects around us. It consists of a mobile website that constructs quirky sentences about the objects it sees through the users camera feed.

As the users camera focuses on an object, a built in machine learning model gives its best guess as to what object it is seeing. Based on this estimate a short query is sent to Googles Suggest API, which in return sends back a list of sentences inspired by the detected object. The poem appears on the screen overlaying the image of the object. It consists of a handful sentences expressing thoughts, questions or comments related to the immediate object as well as the bigger world outside it.

Google Suggest

On average more than 3.5 billion daily searches are performed on Google around the world. Google Suggest works by attempting to complete users search queries by providing a list of

the sentences most searched for on Google. A number of other factors such as location, search history and trending terms also affect the algorithm and determine the order of suggestions. Finally censorship prevents (at least to some extent) hateful, pornographic, piracy-related and some other controversial queries to be suggested.

Hence Poems about things provides a unique insight into the aggregated human behind common Google search queries, reflecting the topics that mankind as a whole seems interested in and the big and small problems we face.

Starting the searches with specific words

In order to make the results from Google Suggest a bit more personal and interesting each detected object is preceded by one or two words starting points from the following list: “my”, “this”, “does”, “is my”, “does my”, “why can”, “which”, “why are”, “what”, “where”, “can”, “help”, “does”, “that”

Trimming the results

To avoid certain types of results I manually created a list of words that were to be removed from the sentences. These included words like: “meme”, “memes”, “lyric”, “lyrics”, “chord”, “chords”, “imdb”, “movie”, “movies”, “cast”, “quotes”


The project is build using Googles Suggest API, the mobileNet image classifier from ml5js and p5js.



Try out the project here