Quick Start Guide

Create a new project

1. At the bottom of the project page, click the “Register new project” button.
   

2. Add the information for the new project, Name, Description and Cover Image, and save.
   

Add project storage

Optional: If you would like to link the ADC with an external cloud server to pull out your images from there.

1. Go your project’s main page. At the bottom of the page there is a menu with several options, click the “Storage” tab.
   

2. Click “Register a new storage”.
   

3. Add the credentials for your external server and the bucket name. If you don’t have a “Public Base URL” you can leave that empty.
   

Add deployments

1. Go your project’s main page and click on on the “Deployments” tab.
   

2. Scroll down to the bottom of the page and click “Register new deployment”.
   

3. Each deployment represents a system in a particular location.

   • Provide a name and desctiption for the deployment.
     
   • And a Location (latitude and longitude).
     
   • Finally, you can link the deployment to your data store or external cloud. For that, you need to select the correct “Data source”, click “Sync” and then add the “Sub-Directory in Data Source” which is the bucket and the key where the data is stored.
     Note: If you haven’t registered your external storage space in the project you will need to do that (section “Add project storage”). However, you can create the deployment and add the images data source later.
     

4. Save.

Process images

1. First, you need to create a collection of the images that you wish to process. Go the project’s main page and click the “Collections” tab.
   

2. Add a new collection.
   

3. Fill the information and save.
   

The suggestion is to create one collection for each week of data collected.
The strategy the MILA team have been using is an "in-fill" method. I make 3 collections:
- Late-night 10 minute intervals (min hour: 2, max hour: 6).
- 1 hour intervals across all nights.
- Everything else (or 10 minute intervals across all nights).
This way you get a good cross section of a lot of moths from all nights & all deployments pretty quickly, without waiting to process the majority of images which are blank. Then the "everything else" job process all of the remaining images over the next week or hours. The jobs filter out images that have already been processed by the same pipeline before they add them to their queue.

4. Once, the collection is created click “Populate”.

5. Move to the “Status” tab.
   

6. Register a new job.
   

7. Fill the job’s information. In the “Source images” field select the collection that you just created and in the "Pipeline" select the classification model that you would like to use, and save.
   

8. Your images are now being processed, wait until the job is completed.