Many of our ADRs deal with the architectural characteristics (ADR 000-013, 017). We selected the following characteristics based on our analysis of Wildlife.AI requirements:
-
cost
-
performance (reasonable performance)
-
scalability (long term scalability)
-
extensibility
-
simplicity
-
testability
-
interoperability
We also considered maintainability.
We used the actor/action approach to break down the domain. We added 3 other actors, but they are only used in very few use cases.
-
Initialize Camera for shipping
-
Create camera inventory records
We added the role of staff who will be needed to facilitate the registration and capabilities of the cameras using the system making sure the serial numbers are correct before shipping.
-
Register User
-
Register Camera
-
Authenticate
-
Turn Camera On/Off (using app)
-
Change Camera Settings (using app)
-
Set Camera location
-
Create Workspace
-
Change security for Workspace
-
Assign content to Workspace
-
Download Video from Camera (offline)
-
Upload Video
-
Select Video for Labeling with criteria
-
Select Video for Training with criteria
-
Download Training Model
-
Share Video
-
Import Shared Video from other user
-
Import Shared Training data
-
Select best frame from video for Expert
-
Review reports for own videos
-
Review reports for geographic region
We added the concept of Workspaces which is like a Project and often the container for the training.
-
Send periodic occurance
-
Send immediate occurance
-
Send camera health alert
With better connectivity and capabilties, the Smart Camera will be able to do more functions like uploading videos that require manual handling at the moment.
Before we offer third party tool / platforms, it is good to look at potential fees. We didn’t see any reason to exclude one particular platform.
Wildlife Insights, TrapTagger and Trapper are Open Source.
Free Community edition for students, universities, and individual developers at companies with less than 15 employee.