Releases: FabianPlum/OmniTrax
OmniTrax Version 1.0.0
The official release of OmniTrax with its corresponding JOSS paper
Deep learning-based multi animal tracking and pose estimation Blender Add-on.
Updates:
- additional sample export functionality
- added Blender usability tips
- new post processing options in example scripts
- updated test runners and requirements for dev work
- major code-base restructuring
- improved exception handling and prompting so things don't crash while leaving you in the dark why
- documentation and installation updates
Notes on CPU-only version
OmniTrax CPU runs on both Windows 10 / 11 as well as Ubuntu systems
If you wish to run inference on CPU-only systems, you must use Blender version 2.92 to match dependencies!
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax GPU is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab, as described in the README. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
OmniTrax Version 0.3.1
The latest version of OmniTrax
Updates:
- improved exception handling so things don't crash while leaving you in the dark why
- documentation and installation updates
- CUDA environment checks
Notes on CPU-only version
OmniTrax CPU runs on both Windows 10 / 11 as well as Ubuntu systems
If you wish to run inference on CPU-only systems, you must use Blender version 2.92 to match dependencies!
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax GPU is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab, as described in the README. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
OmniTrax Version 0.3.0
The latest version of OmniTrax now finally runs under Ubuntu
(well, at least the CPU version does)
Notes on CPU-only version
OmniTrax CPU runs on both Windows 10 / 11 as well as Ubuntu systems
If you wish to run inference on CPU-only systems, you must use Blender version 2.92 instead to match dependencies!
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax GPU is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab, as described in the README. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
OmniTrax Version 0.2.3
The latest version of OmniTrax fixes prior issues relating to masking and yolo path handling (resolving issues #3 #4 #5 )
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab, as described in the README. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
Notes on CPU-only version
If you wish to run inference on CPU-only systems, you must use Blender version 2.9.2 instead to match dependencies!
OmniTrax Version 0.2.2
The latest version of OmniTrax now adds support for footage masking and advanced sample export (see tutorial-tracking for details). Users can now specify in which regions they want to run inference, to, e.g. exclude reflections or animals outside the region of interest. Advanced sample export can be used to extract cropped image patches of individual animals for further analysis or to inform the training of other specialised networks.
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab, as described in the README. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
Notes on CPU-only version
If you wish to run inference on CPU-only systems, you must use Blender version 2.9.2 instead to match dependencies!
OmniTrax Version 0.2.1
The latest release includes a few minor usability updates and improved exception handling for trained YOLO and DLC-live networks. You can now configure the detection (YOLO) network input resolution within OmniTrax directly. OmniTrax will now automatically create .data as well as .names files for YOLO networks based on the supplied .cfg (config) files, in case they are not provided or formatted incorrectly.
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab, as described in the README. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
Notes on CPU-only version
If you wish to run inference on CPU-only systems, you must use Blender version 2.9.2 instead to match dependencies!
OmniTrax Version 0.2.0
Lots of small DeepLabCut-Live improvements, including running inference on full-frames, thus no longer requiring prior tracking. Optionally, for multi-animal tracking bounding boxes can be defined either automatically through yolo-based buffer-and-recover- , blender internal- or manual tracking. Additionally, this latest version includes various bug-fixes as well as unit-tests to improve mainability.
GPU inference compatibility with the latest release of Blender LTS 3.3!
OmniTrax is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version [recommended]
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
Notes on CPU-only version
If you wish to run inference on CPU-only systems, you must use Blender version 2.9.2 instead to match dependencies!
OmniTrax Version 0.1.3
Includes improved tracking from previous states, faster and more robust track transfer, building skeletons from DLC config files, improved package installation and start-up checks, a few bug fixes, and GPU compatibility with the latest release of Blender LTS 3.3!
OmniTrax is built for Windows 10 / 11.
Simply install the omni_trax.zip file from within Blender in the Addon tab. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Notes on GPU version
Install the correct CUDA & cudNN versions (11.2 and 8.1 respectively) mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs with compute capability of 6.1 or higher.
Notes on CPU-only version
If you wish to run inference on CPU-only systems, you must use Blender version 2.9.2 instead to match dependencies!
OmniTrax Version 0.1.2
Including lots of minor updates, this latest release of OmniTrax works in Blender LTS 3.3!
Simply install the omni_trax.zip file from within Blender in the Addon tab. No need to unpack the file, but make sure to run Blender in Administrator mode for the installation of this addon as additional python packages will be installed automatically.
Install the correct CUDA & cudNN versions mentioned in the readme for GPU support. The DLLs have been built specifically to run on machines with dedicated NVIDIA GPUs.
Important
If you wish to run inference on CPU-only systems, you must explicitly install Blender version 2.9.2 to match dependencies!
OmniTrax Version 0.1.1
This newest release includes new DLLs for both CPU and GPU support to start using OmniTrax on more machines. You can now select your computation device for the addon from a dropdown menu. Make sure to install the correct CUDA & cudNN versions mentioned in the readme for GPU support.
Believe me, if you do have a CUDA compatible GPU with sufficient VRAM, you really want to use that over a CPU - The inference speed on GPUs is easily 10 times higher (or more) for detection-based multi-animal tracking.
IMPORTANT
This release is explicitly built for Blender version 2.9.2 to match dependencies!