Screenshot 2025-02-10 153057
The current hardware setup for the development and testing of the Wildlife Watcher.
The Wildlife Watcher team is making big strides as we gear up for testing! From hardware assembly to software refinements and data science tools, everything is coming together. With field testing underway, we’re one step closer to deploying this smart camera for species detection.
We’re already halfway through Q1, and the Wildlife Watcher team is busier than ever! A lot of exciting progress is happening across all workstreams as we prepare the hardware, software, and data science for comprehensive testing.
The Wildlife Watcher is a smart camera designed to detect and identify a wide range of species in the wild. It consists of a camera with enhanced motion detection, built-in AI for species identification, and illumination for low-light conditions. The system also includes a mobile app for basic operations and a desktop app for managing machine learning models.
The project is rolled out in two phases, each with a different Minimum Viable Product (MVP) requirements that target specific use cases. The first focuses on detecting threatened geckos, while the second focuses on identifying rodents for detecting invasive species. We’re making great progress toward delivering the technology needed for the first use case.
Hardware Setup
On the hardware side, multiple printed circuit boards (PCBs) have been successfully assembled, and our proposed solution has been reviewed by our chip/semiconductor partner. The board integrates the Himax HX6538 processor, featuring an internal neural network accelerator, along with a communications module supporting both Bluetooth and LoRaWAN. It connects to a Raspberry Pi camera and includes options for a radar-based motion detection, previously used in the Wētā Watcher, as well as an embedded optical motion sensor.
The full setup—board, camera, motion sensor, illumination, battery, and temporary casing—is currently undergoing field testing. These tests will ensure the system functions as expected, while also providing insight into power consumption and image quality. Since our first MVP focuses on gecko detection, achieving a balance between long battery life and high-quality, illuminated images is key.
A structured testing framework has been established, covering camera functionality, connectivity, image quality, power management, sensor accuracy, and data transmission.
Software Progress
On the software side, we’ve chosen a simpler, more efficient database solution, which is currently being set up and tested. Early results are promising, and we aim to begin mobile app testing in the coming weeks.
From a data science perspective, the team has developed a lightweight annotation tool that allows EXIF data to be easily exported and modified. This will simplify testing and metadata validation. Additionally, we’re finalising the deployment folder structure and naming conventions, ensuring that all deployment data from the mobile application is securely stored in an organised and readable format.
What’s Next?
With hardware testing underway, software refinement in progress, and data science tools taking shape, we’re moving closer to bringing Wildlife Watcher to life. Over the next few weeks, our focus will be on field testing, mobile app integration, and refining our machine learning models.
Stay tuned for more updates, and if you’re interested in getting involved or learning more, feel free to reach out—we’d love to hear from you!
