MediaPipe Hands is a real-time hand tracking and gesture recognition framework developed by Google, based on deep learning. This model can detect and track hands using a single RGB camera, identifying 21 key points of the hand and fingers. MediaPipe Hands employs a lightweight convolutional neural network, allowing it to achieve high-precision gesture recognition and hand tracking with low latency. The algorithm first uses a palm detector to locate the hand, followed by refining the position of key hand points. Its efficiency makes MediaPipe Hands widely used in applications such as virtual reality, gesture control, and augmented reality, providing robust support for real-time interaction systems.

Source model

• Input shape: [1x3x256x256], [1x3x256x256]
• Number of parameters:1.76M, 2.01M
• Model size:7.11MB, 8.09MB
• Output shape: [1x2944x18, 1x2944x1], [1,1,1x21x3]

Source model repository: MediaPipe-Hand

When the user has fine-tuned the source model, the model conversion process must be performed again.

Users can refer to either of the following two methods to complete the model conversion:

• Using AIMO for model conversion: Click Model Conversion Reference in the Performance Reference section on the right to view the conversion steps.

• Using Qualcomm QNN for model conversion: Please refer to the Qualcomm QNN Documentation.

The model performance benchmarks and example code provided by Model Farm are all implemented based on the APLUX AidLite SDK.

For models in .bin format, you can use either of the following two inference engines to run inference on Qualcomm chips:

Inference using APLUX AidLite: please refer to the APLUX AidLite Developer Documentation

Inference using Qualcomm QNN: Please refer to the Qualcomm QNN Documentation

Inference Example Code

The inference example code is implemented using the AidLite SDK.

Click Model & Code to download the model files and the inference code package. The file structure is as follows:

/model_farm_{model_name}_aidlite
    
    |__ models # folder where model files are stored

    |__ python # aidlite python model inference example

    |__ cpp # aidlite cpp model inference example

    |__ README.md