Model ResNet

This function block allows users to utilize various ResNet models for image classification tasks. It provides different options for model selection and input configurations.

📥 Inputs

Choose Folder 2D Connect to an input block that provides image data ready for processing through the ResNet model.

📤 Outputs

The output of this function block is the corresponding model ready to be evaluated on input images.

🕹️ Controls

Model Type A dropdown menu that allows users to select the ResNet variant to use (e.g., ResNet50, ResNet101, etc.).

Input Size A text input to specify the desired input image size for the model. It should be an integer value that meets the minimum requirement (32 pixels).

Pooling A dropdown menu to select the pooling type to be used in the model (Maximum, Average, or None).

🎨 Features

Multiple Model Options Offers a variety of ResNet architectures to choose from, catering to different needs based on performance and complexity.

Dynamic Input Size User-defined input size allows flexibility based on the input images being used.

Pooling Options Users can decide on the pooling layer to be used in the architecture, which can affect the model's classification performance.

📝 Usage Instructions

1. Connect Input: Link this block to an appropriate input block that provides images for processing, such as a folder of images.

2. Select Model Type: Choose the desired ResNet model variant from the Model Type dropdown.

3. Specify Input Size: Enter an integer value in the Input Size field (minimum size is 32).

4. Select Pooling Method: Choose a pooling method from the Pooling dropdown to define how the pooling layers will be handled.

5. Evaluate Model: Run the block, and it will output the ResNet model configured with specified parameters for classification tasks.

📊 Evaluation

Upon evaluating the block, it will return a configured ResNet model prepared for image classification based on the connected image input.

💡 Tips and Tricks

Handling Different Image Sizes

Ensure your input images match the specified input size. Resizing images beforehand may improve model performance and prevent errors.

Choosing the Right Pooling Method

Experiment with different pooling options based on your dataset. For high-resolution images, using Average pooling may help retain relevant features.

Verifying Input Colors

The model works with RGB images only. Ensure that your input data is formatted correctly. You could use an Image Color Match function block to confirm image channels before processing.

🛠️ Troubleshooting

Invalid Input Size Error

If you encounter an input size error, ensure the input provided is an integer value greater than or equal to 32. If the error persists, verify your entered input.

RGB Image Requirement

If the model fails, ensure that the input images are in the correct format (RGB). Any grayscale or different format images will lead to an error.