Pose detection tf-life yolov8 - using int8 weights. about ultralytics HOT 2 OPEN

👋 Hello @adriaanslechten, thank you for your interest in Ultralytics YOLOv8 🚀! We recommend a visit to the Docs for new users where you can find many Python and CLI usage examples and where many of the most common questions may already be answered.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset image examples and training logs, and verify you are following our Tips for Best Training Results.

Join the vibrant Ultralytics Discord 🎧 community for real-time conversations and collaborations. This platform offers a perfect space to inquire, showcase your work, and connect with fellow Ultralytics users.

Install

Pip install the ultralytics package including all requirements in a Python>=3.8 environment with PyTorch>=1.8.

pip install ultralytics

Environments

YOLOv8 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all Ultralytics CI tests are currently passing. CI tests verify correct operation of all YOLOv8 Modes and Tasks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

from ultralytics.

@adriaanslechten hello,

Thank you for reaching out and providing a detailed description of your issue along with the code snippet. Let's work through this together.

First, let's ensure that you are using the latest versions of torch and ultralytics. If not, please upgrade your packages and try again:

pip install --upgrade torch ultralytics

Regarding your code, it looks like you are correctly setting up the model and processing the frames. However, there are a few points to consider:

1. Model Input Size: Ensure that the input size of the frame matches the size used during the model export. If you exported the model with an input size of 128, you should resize your frames to 128x128 before passing them to the model.

2. Output Shape: Verify that the output shape of the model matches your expectations. The output shape should be consistent with the model's architecture and the number of keypoints it predicts.

3. Keypoint Extraction: Ensure that the keypoint extraction logic correctly maps the model's output to the keypoints. Each keypoint should have its coordinates and confidence score extracted accurately.

Here's a refined version of your code with these considerations:

const model = useTensorflowModel(require('../assets/yolov8n-pose_int8.tflite'));
const keyPoints = [
  'NOSE', 'LEFT_EYE', 'RIGHT_EYE', 'LEFT_EAR', 'RIGHT_EAR',
  'LEFT_SHOULDER', 'RIGHT_SHOULDER', 'LEFT_ELBOW', 'RIGHT_ELBOW',
  'LEFT_WRIST', 'RIGHT_WRIST', 'LEFT_HIP', 'RIGHT_HIP',
  'LEFT_KNEE', 'RIGHT_KNEE', 'LEFT_ANKLE', 'RIGHT_ANKLE',
];
type KeyPoint = {
  x: number;
  y: number;
  confidence: number;
  pose: (typeof keyPoints)[number];
};

const frameProcessor = useFrameProcessor(frame => {
  'worklet';
  const resizedFrame = resize(frame, 128, 128); // Ensure this matches the export size

  const outputShape = { batchSize: 1, numberOfDetections: 56, valuesPerDetection: 336 };
  const modelOutput = model.model.runSync([resizedFrame])[0];

  if (modelOutput.length !== outputShape.numberOfDetections * outputShape.valuesPerDetection) {
    console.error('Unexpected model output length:', modelOutput.length);
    return;
  }

  const detectionValues = modelOutput.slice(0, outputShape.valuesPerDetection);
  const detectionsToDraw = [];

  const boundingBox = {
    x: detectionValues[0],
    y: detectionValues[1],
    width: detectionValues[2],
    height: detectionValues[3],
  };

  const detectionConfidence = detectionValues[4];
  const keyPointsData = {} as any;

  for (let keyPointIndex = 0; keyPointIndex < keyPoints.length; keyPointIndex++) {
    const keyPointValueIndex = 5 + keyPointIndex * 3;
    const keyPointX = detectionValues[keyPointValueIndex];
    const keyPointY = detectionValues[keyPointValueIndex + 1];
    const keyPointConfidence = detectionValues[keyPointValueIndex + 2];
    const pose = keyPoints[keyPointIndex];

    keyPointsData[pose] = {
      x: keyPointX * frame.width,
      y: keyPointY * frame.height,
      confidence: keyPointConfidence,
      pose: pose,
    };
  }

  const detectionResult = {
    boundingBox,
    confidence: detectionConfidence,
    keyPoints: keyPointsData as KeyPoint[],
  };

  detectionsToDraw.push(detectionResult);

  Object.values(detectionResult.keyPoints).forEach((keyPoint: KeyPoint) => {
    if (keyPoint.confidence > 0.4) {
      console.log('Keypoint:', keyPoint.x, keyPoint.y);
    }
  });
}, [model]);

If the issue persists, please provide a minimum reproducible example so we can investigate further. You can find more details on creating a minimum reproducible example here.

Feel free to reach out if you have any more questions or need further assistance. 😊

from ultralytics.