Age-Gender Recognition

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

0 Likes

Model Details

Model Description

Age-Gender Recognition is a fully convolutional neural network model designed to estimate a person's age and gender from their face features. For optimal results, it should be used in conjunction with a face detector to input only the cropped face image. Please note that the model is trained to recognize the ages of individuals between 18 and 75 years old.

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

Photo by Daniel Xavier from

Training Details

Training Data

Internal datataset of (20k) diverse faces.

Testing Details

Metrics

The model was tested on internal dataset of divers faces. Average age estimation error was 6.99 years and gender accuracy was 95.80%.

Technical Specifications

Input/Output Details

Input:
- Name: image
  - Info: NCHW BGR un-normalized image of a face cropout
Output 1:
- Name: age_conv3
  - Info: estimate of age
Output 2:
- Name: prob
  - Info: probability of female or male

Model Architecture

Fully convolutional model with 8 convolutional layers.

Throughput

Model variant: age-gender-recognition:62x62

• Input shape: [1, 3, 62, 62] • Output shapes: [[1, 1, 1, 1], [1, 2, 1, 1]]

• Params (M): 2.138 • GFLOPs: 0.048

Platform	Precision	Throughput (infs/sec)	Power Consumption (W)
RVC2	FP16	721.43	N/A
RVC4	INT8	687.67	2.64

* 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 a 50-image subset of the dataset followed by utilizing YuNet model to make face cropouts.

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/age-gender-recognition:62x62"
)

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

Inspect model head(s):

RegressionParser that outputs message (detected age).
ClassificationParser that outputs message (detected gender).

The model is multi-headed. You can set up the queues as follows:

regression_parser_output_queue = nn.getOutput(0).createOutputQueue()
classification_parser_output_queue = nn.getOutput(1).createOutputQueue()

Get parsed output(s):

while pipeline.isRuning():
    regression_parser_output: Predictions = regression_parser_output_queue.get()
    classification_parser_output: Classifications = classification_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 consisting of a face detection model and an age recognition model. It automatically downloads the models, creates a DepthAI pipeline, runs the inference, and displays the results using our DepthAI visualizer tool.

To try it out, run:

python3 main.py

Gender classification and age regression model.
License	Apache 2.0 Commercial use
Downloads	738
Tasks	Classification
Model Types	ONNX

Model Variants

Name	Version	Available For	Created At	Deploy
		RVC2, RVC4	9 months ago