Use MoviNet for binary classification problem

[shahmirshabbir]

Prerequisites

Please answer the following question for yourself before submitting an issue.

• I checked to make sure that this feature has not been requested already.

1. The entire URL of the file you are using

https://github.com/tensorflow/models/tree/master/official/...

2. Describe the feature you request

A clear and concise description of what you want to happen.

3. Additional context

Add any other context about the feature request here.

4. Are you willing to contribute it? (Yes or No)

[shahmirshabbir]

Can someone tell how to use MoViNet for binary classification problem

[malla456]

Hi @shahmirshabbir ,

To use MoViNet for binary classification, format your video data into 5D tensors (batch_size, num_frames, height, width, channels). Then, load a pre-trained MoViNet base model and exclude its top classification layer, which is normally set up for multi categories. Initially "freeze" this base model to have its pre-learned ability to recognize visual features.
Next, append a custom classification head. To do this effectively, add a Global 3D Average Pooling layer to flatten the complex video features, followed by a Dropout layer  to prevent the model from overfitting your data. End the network with a single Dense(1) layer using a sigmoid activation function. This forces the model to output a single probability score between 0.0 and 1.0.
Finally, compile the model using BinaryCrossentropy as your loss function.

Thank you.