首页 > > 详细

代写ECS 174代做R程序、R程序代做、调试R作业

ECS 174: Intro to Computer Vision, Spring 2020 
Problem Set 3 
 
Instructor: Yong Jae Lee () 
Instructor: Krishna Kumar Singh () 
TA: Haotian Liu () 
TA: Utkarsh Ojha () 
TA: Yuheng Li () 
 
 
Due: Monday, June 1st, 11:59 PM 
 
 
Instructions 
1. Answer sheets must be submitted on Canvas. Hard copies will not be accepted. 
 
2. Please submit your answer sheet containing the written answers in a file named: 
FirstName_LastName_PS3.pdf. 
 
3. Please submit your code and input /output images in a zip file named: 
FirstName_LastName_PS3.zip. Please do not create subdirectories within the main directory. 
 
4. You may complete the assignment individually or with a partner (i.e., maximum group of 2 
people). If you worked with a partner, provide the name of your partner. We will be using 
MOSS to check instances of plagiarism/cheating. 
 
5. For the implementation questions, make sure your code is documented, is bug-free, and works 
out of the box. Please be sure to submit all main and helper functions. Be sure to not 
include absolute paths. Points will be deducted if your code does not run out of the box. 
 
6. If plots are required, you must include them in your answer sheet (pdf) and your code must 
display them when run. Points will be deducted for not following this protocol. 
 
 
1 Short answer problems [10 points] 
 
1. What exactly does the value recorded in a single dimension of a SIFT keypoint descriptor signify? 
 
 
 
2. When performing interest point detection with the Laplacian of Gaussian, how would results differ 
if we were to (a) take any positions that are local maxima in scale-space, or (b) take any positions 
whose filter response exceeds a threshold? Specifically, what is the impact on repeatability or 
distinctiveness of the resulting interest points? 
 
 
2 Programming: Video search with bag of visual words [90 points] 
 
 
For this problem, you will implement a video search method to retrieve relevant frames from a video 
based on the features in a query region selected from some frame. We are providing the image data and 
some starter code for this assignment. 
 
 
Provided data 
 
You can access pre-computed SIFT features here: 
https://drive.google.com/file/d/10yk7tvDfmge9fEVm2XbwAmaIRL9R7clK/view?usp=sharing 
 
The associated images are stored here: 
https://ucdavis.box.com/s/ylxih5tgwja1azx78jkc0d5awcxla71m 
 
Please note the data takes about 6 GB. Each .mat file in the provided SIFT data corresponds to a single 
image, and contains the following variables, where n is the number of detected SIFT features in that image: 
 
descriptors nx128 double // SIFT vectors as rows 
imname 1x57 char // name of image file that goes with this data 
numfeats 1x1 double // number of detected features 
orients nx1 double // orientations of the patches 
positions nx2 double // positions of the patch centers 
scales nx1 double // scales of the patches 
 
Provided code 
 
The following are the provided code files. You are not required to use any of these functions, but you will 
probably find them helpful. You can access the code here: 
https://ucdavis.box.com/s/cll544a6gq4zaqgf6emn9uf3cq5gwy51 
 
• loadDataExample.m: Run this first and make sure you understand the data format. It is a 
script that shows a loop of data files, and how to access each SIFT descriptor. It also shows 
how to use some of the other functions below. 
 
• displaySIFTPatches.m: given SIFT descriptor info, it draws the patches on top of an image 
 
• getPatchFromSIFTParameters.m: given SIFT descriptor info, it extracts the image patch 
itself and returns as a single image 
 
• selectRegion.m: given an image and list of feature positions, it allows a user to draw a polygon 
showing a region of interest, and then returns the indices within the list of positions that fell 
within the polygon. 
 
• dist2.m: a fast implementation of computing pairwise distances between two matrices for 
which each row is a data point 
 
• kmeansML.m: a faster k-means implementation that takes the data points as columns 
What to implement and discuss in the write-up 
 
Write one script for each of the following (along with any helper functions you find useful), and in your pdf 
writeup report on the results, explain, and show images where appropriate. Your code must access the frames 
and the SIFT features from subfolders called ‘frames’ and ‘sift’, respectively, in your main working 
directory. 
 
 
1. Raw descriptor matching [20 pts]: Allow a user to select a region of interest (see provided 
selectRegion.m) in one frame, and then match descriptors in that region to descriptors in 
the second image based on Euclidean distance in SIFT space. Display the selected region of interest 
in the first image (a polygon), and the matched features in the second image, something like the 
below example. Use the two images and associated features in the provided file 
twoFrameData.mat (in the zip file) to demonstrate. Note, no visual vocabulary should be 
used for this one. Name your script raw_descriptor_matches.m 
 
 
 
2. Visualizing the vocabulary [25 pts]: Build a visual vocabulary. Display example image patches 
associated with two of the visual words. Choose two words that are distinct to illustrate what the 
different words are capturing, and display enough patch examples so the word content is evident 
(25 patches per word displayed). See provided helper function 
getPatchFromSIFTParameters.m. Explain what you see. Name your script 
visualize_vocabulary.m. Please submit your visual words in a file called kMeans.mat. 
This file should contain a matrix of size kx128 called kMeans. 
 
3. Full frame queries [25 pts]: After testing your code for bag-of-words visual search, choose 3 
different frames from the entire video dataset to serve as queries. Display each query frame and 
its M=5 most similar frames (in rank order) based on the normalized scalar product between 
their bag of words histograms. Explain the results. Name your script 
full_frame_queries.m 
 
4. Region queries [20 pts]: Select your favorite query regions from 3 frames of your choice (which 
may be different than those used above) to demonstrate the retrieved frames when only a portion 
of the SIFT descriptors are used to form a bag of words. Try to include example(s) where 
the same object is found in the most similar M frames but amidst different objects or 
backgrounds, and also include a failure case. Display each query region (marked in the frame 
as a polygon) and its M=5 most similar frames. Explain the results, including possible reasons 
for the failure cases. Name your script region_queries.m 
 
 
 
 
 
Tips: overview of framework requirements 
 
The basic framework will require these components: 
 
• Compute nearest raw SIFT descriptors. Use the Euclidean distance between SIFT descriptors 
to determine which are nearest among two images’ descriptors. That is, “match” features from 
one image to the other, without quantizing to visual words. 
 
• Form a visual vocabulary. Cluster a large, representative random sample of SIFT descriptors 
from some portion of the frames using k-means. Let the k centers be the visual words. The 
value of k is a free parameter; for this data something like k=1500 should work, but feel free 
to play with this parameter [see Matlab’s kmeans function, or provided kmeansML.m code]. 
Note: you may run out of memory if you use all the provided SIFT descriptors to build the 
vocabulary. 
 
• Map a raw SIFT descriptor to its visual word. The raw descriptor is assigned to the nearest 
visual word. [see provided dist2.m code for fast distance computations] 
 
• Map an image’s features into its bag-of-words histogram. The histogram for image I j is a k- 
dimensional vector: F (I j ) = [ freq1,j , freq2,j , … , freqk,j], where each entry freqi,j counts the 
number of occurrences of the i-th visual word in that image, and k is the number of total words 
in the vocabulary. In other words, a single image’s list of n SIFT descriptors yields a k- 
dimensional bag of words histogram. [Matlab’s histc is a useful function] 
 
• Compute similarity scores. Compare two bag-of-words histograms using the normalized 
scalar product. 
 
• Sort the similarity scores between a query histogram and the histograms associated with the 
rest of the images in the video. Pull up the images associated with the M most similar examples. 
[see Matlab’s sort function] 
 
• Form a query from a region within a frame. Select a polygonal region interactively with the 
mouse, and compute a bag of words histogram from only the SIFT descriptors that fall within 
that region. [see provided selectRegion.m code] 
 
• There may be some frames (e.g., all black) in which no features are detected and hence no 
descriptors are available. If so, you will need to ignore those frames (e.g., using an if statement). 
 
 
 
 
 
3 OPTIONAL: Extra credit (10 points) 
 
• Stop list and tf-idf. Implement a stop list to ignore very common words, and apply tf-idf weighting 
to the bags of words. Discuss and create an experiment to illustrate the impact on your results. 
 
 
 
 
 
 
 
 
 
 
联系我们
  • QQ:99515681
  • 邮箱:99515681@qq.com
  • 工作时间:8:00-23:00
  • 微信:codinghelp2
热点文章
程序代写更多图片

联系我们 - QQ: 99515681 微信:codinghelp2
© 2014 www.7daixie.com
程序代写网!