khushboo-agarwal's repositories
Optical-Flow
Implement Lucas-Kanade optical flow estimation, and test it for the two-frame data sets provided in Python from scratch
Action-Recognition
recognize actions from videos using machine learning classifier(s) and suitable features. You will use UCF sports action data set here http://crcv.ucf.edu/data/ucf_sports_actions.zip. UCF Sports dataset consists of a set of actions collected from various sports which are typically featured on broadcast television channels such as the BBC and ESPN. The video sequences were obtained from a wide range of stock footage websites including BBC Motion gallery and GettyImages. The dataset includes a total of 150 sequences with the resolution of 720 x 480. The collection represents a natural pool of actions featured in a wide range of scenes and viewpoints. By releasing the data set we hope to encourage further research into this class of action recognition in unconstrained environments. Since its introduction, the dataset has been used for numerous applications such as: action recognition, action localization, and saliency detection. The dataset includes the following 10 actions. The figure above shows the a sample frame of all ten actions, along with their bounding box annotations of the humans shown in yellow.
Susan-Corner-Detection
SUSAN corner detection algorithms in Python from scratch
Corner-Detection
[1 pts] Implement corner detection algorithm based on Hessian matrix (H) computation. Note that Hessian matrix is defined for a given image I at a pixel p as Ixx(p) Ixy(p) H1(p) = Ixy(p) Iyy(p) , (0.1) such that eigen-decomposition (spectral decomposition) of this matrix yields two eigenvalues as: λ1 and λ2. If both λ1,λ2 are large, we are at a corner. Provide the detected corners in the resulting output images in color. [1 pt] Implement Harris Corner Detection algorithm for the same input images you used in previous question. Rather than considering the Hessian of the original image I (i.e. second-order derivatives), we use the first-order derivatives of the smoothed version L(p, σ) for some Gaussian filter with standard deviation σ > 0. Note that you need to construct the following matrix for every pixel p,  L2x(p, σ) Lx(p, σ)Ly(p, σ) Lx(p, σ)Ly(p, σ) L2y(p, σ) , (0.2) H2(p, σ) = where L is obtained after smoothing I with Gaussian filter G. Now, instead of calculating those eigenvalues we computed in previous question, we will consider the cornerness measure as Cornerness(p, σ, α) = Det(H2) − α.Tr(H2), (0.3) where Det and Tr indicate the determinant and trace of the matrix H2, respectively. Please use non-negative α ≈ 1/25 as a starting value and try to optimize it by trying different values and comment about it. Provide the detected corners in the resulting output image in color. [1 pt] In the previous question, replace cornerness measure with the following: Cornerness(p, σ, α) = λ1λ2 − α(λ1 + λ2), (0.4) and determine the efficiency of this system and the system in the previous question by measuring and reporting the time. You are supposed to get the same results in accuracy but different results in efficiency.
Entropy-Thresholding
The use of Entropy information for mapping image intensity values into two classes: white (foreground) and black (background). (Python Implementation)
algorithms-divide-and-conquer
This repository basically deals with all the codes taught in the Algorithms MOOC by Stanford University on Coursera in Python
Canny-Edge-Detection
Python Implementation of Canny Edge Detection Implementation from scratch
Coding_Practice
A repository to keep all my coding practice
datasciencecoursera
This is a repository to hold my project during the course of Data Science Specialization from Coursera
datasharing
The Leek group guide to data sharing
DeepCons
Understanding sequence conservation with deep learning
Gaussian-Mean-Shift-Clustering
Implement mean-shift clustering algorithm for the segmentation of the following RGB image: “input3.jpg". Set the parameters (hs: spatial resolution, hr: range resolution) according to the properties of the attached image. Briefly mention optimality of why and how you decide those parameters along with the output. There is no ground truth available for this image as this question is intended to be a blind-competitions among attendants. Best segmentation method will be decided and announced based TP rate, FP rate, and F-score (use isotropic Gaussian kernel). In the pseudo-code, p(n|x) is posterior probability, and x’s update means difference of x’s values between consecutive iterations.
PA0-Python-coding
Some basic python programming questions
Stem_localization
Identify the no. of stems and stem location in images taken from rover
tensorflow
Computation using data flow graphs for scalable machine learning