v0.4.0

What's Changed

• 🚀 Persist EXIF Data in Image Exports by @bencevans in #130

Some Changes to the Testing and Release Workflows

• Fix Testing Builds by @bencevans in #124
• Disable macOS actions release build by @bencevans in #116

Dependency Upgrades

• Bump tauri from 1.2.3 to 1.2.4 in /src-tauri by @dependabot in #122
• Bump bumpalo from 3.10.0 to 3.12.0 in /src-tauri by @dependabot in #123
• Bump web-vitals from 3.1.0 to 3.1.1 by @dependabot in #117
• Bump @tauri-apps/cli from 1.2.2 to 1.2.3 by @dependabot in #120
• Bump react-spinners from 0.13.7 to 0.13.8 by @dependabot in #121
• Bump tokio from 1.24.1 to 1.25.0 in /src-tauri by @dependabot in #125
• Bump serde_json from 1.0.91 to 1.0.92 in /src-tauri by @dependabot in #126
• Bump serde_json from 1.0.92 to 1.0.93 in /src-tauri by @dependabot in #128

Full Changelog: v0.3.0...app-v0.4.0

Installation Notes

Installers can be found under the Assets below. The file required is based on the operating system. This can be identified with:

• Windows (x64) - …x64_en-US.msi
• Windows (x64) with CUDA Acceleration …x64_en_US-cu117.msi
• Linux (x64) - …amd64.deb or …amd64.AppImage
• macOS Intel - …x64.dmg
• macOS M1/M2 - ...aarch64.dmg

Windows users, it’s recommended not to use the CUDA Acceleration version unless the computer already has the NVIDIA CUDA Toolkit 11 installed and a compatible graphics card.

macOS users, if unsure if your computer is an Intel or an M1/M2 device, this can be found in About This Mac.

v0.3.0

Installers can be found under the Assets below. The file required is based on the operating system. This can be identified with:

• Windows (x64) - …x64_en-US.msi
• Windows (x64) with CUDA Acceleration …x64_en_US-cu117.msi
• Linux (x64) - …amd64.deb or …amd64.AppImage
• macOS Intel - …x64.dmg
• macOS M1/M2 - ...aarch64.dmg

Windows users, it’s recommended not to use the CUDA Acceleration version unless the computer already has the NVIDIA CUDA Toolkit 11 installed and a compatible graphics card.

macOS users, if unsure if your computer is an Intel or an M1/M2 device, this can be found in About This Mac.

What's Changed

• Bump tokio from 1.20.1 to 1.24.1 in /src-tauri by @dependabot in #112
• Hide Console on Windows by @bencevans in #109
• Installation instructions in README by @bencevans in #111
• Base ETA prediction on the median to mitigate outliers by @bencevans in #113
• Enable checking for updates by @bencevans in #114
• Signed and Notarized macOS builds to prevent issues mentioned in #108

Full Changelog: v0.2.0...app-v0.3.0

v0.2.0

See the assets to download this version and install.

macOS: If you run into issues running on macOS, please see #108 (comment) for the related workaround, else, feel free to get in touch!

Included in v0.2.0

Multi-Platform and NVIDIA CUDA Acceleration Builds

We now have builds available for all intended platforms:

• Windows (x64) - …x64_en-US.msi
• Windows (x64) with CUDA Acceleration …x64_en_US-cu117.msi
• Linux (x64) - …amd64.deb or …amd64.AppImage
• macOS Intel - …x64.dmg
• macOS M1/M2 - ...aarch64.dmg

The latest builds can be downloaded from https://github.com/bencevans/camtrap-detector/releases under assets. The …’s above match the files under assets to help identify which version is desired.

🙋The CUDA build needs testing on another computer. It’s only been tested on the machine it was built on so far.

Configurable Confidence Threshold

There’s now a cog on the first screen, clicking this will provide an option to change the minimum confidence. By default it’s set to 0.1.

Non-Max Suppression to match MegaDetector Scripts

Non-Max Suppression is now set the same as the MegaDetector Scripts. When the MegaDetector model executes, it outputs a bunch of detections (bounding box and class), sometimes there will be detections that have a large overlap with each other (likely over the same thing …animal/human/vehicle). Non-Max Suppression is a post-processing step that reduces detections that have a large overlap down to one single detection. This was set at 10%, it’s now been fixed to use the same as the MegaDetector Scripts at 45%.

Spot detections in images exports easier

Changed from the fluorescent green colour which was often hard to spot in grassland to show animal detections to white.

Straight onto the next run

There’s now a button to start a new run without needing to restart the application.

Faster Application Start Up

The application now boots faster as it’s not loading the detection model into memory at startup.

Improved the time remaining estimation

Bases the prediction on the intervals between the last n completed inferences.

Full Changelog: v0.1.0...v0.2.0