0

I cannot seem to diagnose any issue in my code yet after 9 epochs of training I get the error in the title.

Here is the code that stores values inside the posenet listener callback

    if (poses.length > 0) {
        pose = poses[0].pose;
        skeleton = poses[0].skeleton;
        let inputs = [];
        if (isCapturing) {
            for (let keypoint of pose.keypoints) {
                let x = keypoint.position.x;
                let y = keypoint.position.y;
                inputs.push(x);
                inputs.push(y);
            }
            if (targetSwitch == 1) { poses1.push(inputs); }
            if (targetSwitch == 2) { poses2.push(inputs); }
        }
    }
}

and here is where i add the data to the neural net

if (key == 't') {
        for (var i = 0; i < poses1.length; i++){
            input = poses1[i];
            output = poses2[i];

            debugArIn.push(input);
            debugArOut.push(output)
            brain.addData(input, output);
        };
        brain.normalizeData();
        //each pose is stored with 34 values (x and y values for each landmark)
        brain.train({epochs:34, batchsize:68}) 
    }

Trying to create a regression model in ml5.js that takes a 2d array of posenet landmark coordinates as inputs and another of the same type as outputs.

50NNY1
  • 1
  • 1
    I think the code provided doesn't show the error. There should be a variable somewhere in there called `e` and it's being called like a function `e()`. Could you provide more code? – denislexic Nov 08 '22 at 02:12
  • Is this wrapped in a `try{}catch(e){}` ? – denislexic Nov 08 '22 at 02:12

0 Answers0