SuperAnimal Landmarker

Our new model ZOO works with DepthAI V3. Find out more in our documentation.

0 Likes

Model Details

Model Description

The DLC SuperAnimal keypoints model is intended to be used for pose estimation of quadruped (animals that have 4 legs) images taken from side view. It has 39 keypoints. Model can estimate the pose of more than 45 different species: from mice, rats, horses, dogs, and cats, to elephants and gazelles. Input image should be already cropped so it only contains the desired animal

Developed by: Mathis Lab
Shared by:
Model type: Computer Vision
License:
Resources for more information:

Training Details

Training Data

Quadruped-80K dataset which is built from over 85,000 images sourced from diverse laboratory settings and in-the-wild data. For more information check the .

Testing Details

Metrics

Evaluation results are taken from .

Dataset	mAP	RMSE
AP-10K	80.11	11.30
AnimalPose	87.03	4.64
Horse-10	95.17	1.15

Technical Specifications

Input/Output Details

Input:
- Name: input
  - Info: NCHW BGR un-normalized image
Output:
- Name: heatmaps
  - Info: Heatmaps for all 39 keypoints. Additional postprocessing is required (see the parser node).

Model Architecture

Please consult for more information on model architecture.

Throughput

Model variant: superanimal-landmarker:256x256

• Input shape: [1, 3, 256, 256] • Output shape: [1, 64, 64, 39]

• Params (M): 28.510 • GFLOPs: 10.268

Platform	Precision	Throughput (infs/sec)	Power Consumption (W)
RVC2	FP16	13.75	N/A
RVC4	INT8	509.90	4.82

* Benchmarked with , using 2 threads (and the DSP runtime in balanced mode for RVC4).

* Parameters and FLOPs are obtained from the package.

Quantization

RVC4 version of the model was quantized using a custom dataset. This was created by taking 70-image subset of dataset.

Utilization

Models converted for RVC Platforms can be used for inference on OAK devices. DepthAI pipelines are used to define the information flow linking the device, inference model, and the output parser (as defined in model head(s)). Below, we present the most crucial utilization steps for the particular model. Please consult the docs for more information.

Install DAIv3 and depthai-nodes libraries:

pip install depthai
pip install depthai-nodes

Define model:

model_description = dai.NNModelDescription(
    "luxonis/superanimal-landmarker:256x256"
)

nn = pipeline.create(ParsingNeuralNetwork).build(
    <CameraNode>, model_description
)

Inspect model head(s):

SuperAnimalParser that outputs message (at most 39 detected keypoints with scores).

Get parsed output(s):

while pipeline.isRuning():
    parser_output: Keypoints = parser_output_queue.get()

Example

You can quickly run the model using our example.

The example demonstrates how to build a 2-stage DepthAI pipeline for detecting animals and estimating their poses. The pipeline consists of object detector and pose estimation model. The example works on both RVC2 and RVC4. For realtime application you will need to use OAK4 cameras.

It automatically downloads the model, creates a DepthAI pipeline, runs the inference, and displays the results using our DepthAI visualizer tool.

To try it out, run:

python3 main.py

Animal landmark model for quadrupeds.
License	GNU General Public License v3.0 Commercial use
Downloads	912
Tasks	Keypoint Detection
Model Types	ONNX

Model Variants

Name	Version	Available For	Created At	Deploy
		RVC2, RVC4	About 1 year ago