Running Classify
To run Mothbot Classify, you just need to go to this special link
The app will still run locally on your computer (nothing gets uploaded to the internet), but it runs in your browser window!
Hackers can get the repo and run the command
bun dev
and you can run it from the scripts too!
You choose your data that has been processed and a species list, and you are set to go!
Choose a Species List
Preliminary Instructions (by Chris Gould)
We haven’t had time to make full documention for this cool new software yet, but the open hardware replicator, Chris Gould, did over on the GOSH forums! They are copied here to help you get started!
After processing my images using the Mothbot program, my original folder named “keyTitira_2026-05-26” now had an additiona folder called “_processed” that contains all the json and image files created by the Mothbot Process program
The next step was to use the Mothbot classify program that runs in your browser.
Before running the program, I had to reorganize my files and include a species list. I made a new folder on my desktop called “Mothbox data”. This contained a folder for my processed photos named “Sieg Hall” (the location of my mothbox), and a folder for my species list named “Species”. It is important that the species folder is named “Species” or the program will not load the species list.
Inside the Sieg Hall folder is the keyTitira_2026-05-26 folder containing my processed photos. In order for the program to work, I had to rename “keyTitira_2026-05-26” to only include the date “2026-05-26”.
Inside the Species folder is the species list (csv) that I used in the processing step.
From here I uploaded the “Mothbox Data” folder to the Mothbot Classify program and classified my patches.
The next step in the process was to parse through the classifications and clusters and manually review them.
By selecting the cluster ID in the top left of a photo, you can highlight all of the photos that are clustered together. Once highlighted, you can accept the classification the program provided (hotkey A) or manually identify the cluster (hotkey D).
There is an option to label a cluster as “ERROR” in the case that the photo contains strange artifacts or in my case photos of the box’s arms.