SarvagyaVaish

### Keybase proof

I hereby claim:

  * I am Sarvagyavaish on github.
  * I am survy (https://keybase.io/survy) on keybase.
  * I have a public key whose fingerprint is A329 DF37 01E3 CCBD CE44  AA48 899A B582 D345 1844

To claim this, I am signing this object:

SarvagyaVaish

import numpy as np
import cv2
import math

def hsv2rgb(h, s, v):
    h = float(h)
    s = float(s)
    v = float(v)
    h60 = h / 60.0
    h60f = math.floor(h60)

SarvagyaVaish

import numpy as np
import cv2
import math

cap = cv2.VideoCapture(0)
ret, frame_large = cap.read()
scale = 0.5
frame = cv2.resize(frame_large, None, fx=scale, fy=scale, interpolation=cv2.INTER_CUBIC)
y, x, c = np.shape(frame)
center = (y/2, x/2)

SarvagyaVaish

TextView title_TextView = (TextView) findViewById(R.id.title);
Paint title_Paint = title_TextView.getPaint();
Paint.FontMetrics title_FontMetrics = title_Paint.getFontMetrics();

TextView subheading_TextView = (TextView) findViewById(R.id.subheading);
Paint subheading_Paint = subheading_TextView.getPaint();
Paint.FontMetrics subheading_FontMetrics = subheading_Paint.getFontMetrics();

Log.d("Survy", "title descent:      " + title_FontMetrics.descent);
Log.d("Survy", "title ascent:       " + title_FontMetrics.ascent);

SarvagyaVaish

What is it?

• an algorithm to solve non-linear least squares minimization problems
• finds a local minima, not necessarily the global minima
• alternatives: Gauss–Newton, gradient-descent

Why use it?

• LM algorithm combines the advantages of gradient-descent and Gauss-Newton
• allows for initial values to be further away from solution than Gauss-Newton
• gradient-descent dominates until a canyon is reached, after which Gauss-Newton takes over

SarvagyaVaish

//
// Create a KDL kinematic Chain.
//
// A Chain is made up of Segments. Each Segment consists of a Joint and a Frame.
// The Joint indicates how the Frame moves - rotation or translation about / along an axis.
//

Chain kdlChain = Chain();

Joint joint1(Joint::None);

SarvagyaVaish

-10.00 -685.80
-9.50 -647.10
-9.00 -602.00
-8.50 -548.90
-8.00 -524.20
-7.50 -490.10
-7.00 -430.60
-6.50 -412.10
-6.00 -345.20
-5.50 -344.10

SarvagyaVaish

struct LMFunctor
{	
    // Compute 'm' errors, one for each data point, for the given parameter values in 'x'
    int operator()(const Eigen::VectorXf &x, Eigen::VectorXf &fvec) const
    {
        // TODO
    }

    // Compute the jacobian of the errors
    int df(const Eigen::VectorXf &x, Eigen::MatrixXf &fjac) const

SarvagyaVaish

// Compute 'm' errors, one for each data point, for the given parameter values in 'x'
int operator()(const Eigen::VectorXf &x, Eigen::VectorXf &fvec) const
{
    // 'x' has dimensions n x 1
    // It contains the current estimates for the parameters.

    // 'fvec' has dimensions m x 1
    // It will contain the error for each data point.

    float aParam = x(0);

SarvagyaVaish

// Compute the jacobian of the functions
int df(const Eigen::VectorXf &x, Eigen::MatrixXf &fjac) const
{
    // 'x' has dimensions n x 1
    // It contains the current estimates for the parameters.

    // 'fjac' has dimensions m x n
    // It will contain the jacobian of the errors, calculated numerically in this case.

    float epsilon;