Action Recognition on Brainy Pi

Action recognition is an exciting field of computer vision that involves identifying human actions in videos or images. In this blog, we will explore how to perform action recognition on Brainy Pi, AI enabled SBC.

To begin with, we need to install the TensorFlow-Hub library by running the “pip install tensorflow-hub” command. We will also import other necessary libraries such as OpenCV, NumPy, and SSL. Next, we define a utility function to fetch videos from the UCF101 dataset, which is a benchmark dataset for action recognition research. We also define another function to open video files using OpenCV.

Install Dependancies

pip install tensorflow-hub

To deploy the code on Brainy Pi, we first need to clone the repository using the “git clone” command. Then, we navigate to the action-recognition directory and run the “python action_recognition.py” command. This will start the action recognition process, and we can see the results in the terminal.

git clone https://github.com/brainypi/BrainyPi-AI-Examples.git
cd BrainyPi-AI-Examples/Tensorflow/action-recognition
python action_recognition.py

Finally, we get the top 5 action predictions, which include “doing a cartwheel,” “cheerleading,” “juggling soccer ball,” “breakdancing,” and “punching bag.” These predictions are based on the sample video that we used for testing the model. We can also modify the code to work with other videos and get their corresponding action predictions.

Conclusion

In conclusion, action recognition is a fascinating field that has several practical applications, such as security surveillance, sports analysis, and human-computer interaction. With the power of Brainy Pi, we can perform action recognition in real-time and create intelligent systems that can understand human actions.

Stay tuned for more such AIoT applications.