Pose Estimation

This function block is used for detecting specific body parts in an image and drawing a skeleton to visualize the estimated posture. It utilizes deep learning models to identify key points associated with human poses.

📥 Inputs

Image Any The input image from which pose estimation will be performed.

Show Skeleton A Boolean input that allows you to toggle the visibility of the skeleton overlay on the image.

📤 Outputs

Image Any The output modified image displaying the detected pose and, if enabled, the skeleton.

Selected Body Part Positions This output provides a mapping of detected body parts and their positions in the image.

🕹️ Controls

Confidence A slider that adjusts the confidence threshold for detecting body parts. Higher values require more certainty from the model to consider a point valid.

Class Names A table view that allows you to select which body parts should be detected and displayed in the output.

🎨 Features

Real-Time Detection This block uses a pre-trained deep learning model to detect body parts in real time and visualize results.

Customizable Detection Users can select specific body parts to monitor, making this tool flexible for various applications.

Skeleton Visualization The detected positions can be connected with lines to form a skeleton, helping in visual understanding of the pose.

📝 Usage Instructions

1. Input Image: Connect an image containing a person or human figure to the Image Any input.

2. Show Skeleton Option: Optionally connect a boolean signal to the Show Skeleton input to enable/disable the skeleton overlay.

3. Adjust Confidence: Use the Confidence slider to adjust the minimum confidence level required for detecting body parts.

4. Select Body Parts: Check or uncheck body parts in the Class Names table to configure which body parts should be detected.

5. Evaluate: Run the block to perform pose estimation. The output will be the modified image and detected body part positions.

📊 Evaluation

When run, this block will output an image displaying the detected body parts and a skeleton if enabled, along with a dictionary of body parts and their positions.

💡 Tips and Tricks

Improving Detection Accuracy

Adjust the Confidence slider to find an optimal balance between detection sensitivity and false positives. A higher value may yield more accurate detections.

Selecting Specific Body Parts

Select only the necessary body part classes you need to analyze; this will simplify the output and focus on the relevant parts of the body.

Preprocessing Input Images

Consider preprocessing input images with contrast adjustment, resizing, or filtering before sending them to this block to enhance detection results.

🛠️ Troubleshooting

No Detections Found

If the block fails to detect any body parts, ensure the input image is clear and well-lit. Lowering the confidence threshold may also help in detecting more points.

Skeleton Not Showing

Make sure the Show Skeleton input is activated and check if any body parts are being detected. If no parts are detected, the skeleton won’t appear.