Daily Scrum

Friday, January 22, 2016

Yesterday: (Last two weeks). Struggled wtih this website. It seems PmWiki has updated likely broken the use of Blog-It for posting. (Issues with formatting, and unable to associate with the blog it groups - via missing post controls).

Took time to go over Faramarz's feedback from the draft document of last fall. There are a lot of things that will need to be re-worked in whatever form it takes next. Started part-time work, and had bereavement issues last week.

Aliasing Issues
My drawers PCD Dataset

No Aliasing Issues!
My drawers PCD Dataset

(Yesterday) Finally worked out the issues causing aliasing on my in-house examples when loaded into SiftFu. Wasn't creating the depth files correctly. They appear to be accurate now, and generate better images. The stationary parts of the scene are not as smooth as desired, but this may be due to the dynamic nature of the in-house examples.

Today: Look at incorporating rigid feature labels for constructing volumetric representation in SiftFu. Visualize the results of this for the different groupings selected. Will need to look at extracting codebase from Matlab, or incorporating a GPU implementation for speed ups (its is awfully slow); as well as RGB colouring of results. Read more...

January 22, 2016, at 04:41 AMerika

Wednesday, January 6, 2016

Yesterday: (Last term and a half). Was tasked with formally writing up the unsuccessful results mid-summer. Summer vacation. Continued writing through the fall, submitting a starter-draft midway through. Numerous suggestions for revisions, though the dominate question is what specifically should be described. Requiring better results before an improved iteration. Reached out (December) to researcher Barnes in Australia who has had success with similar work through student Perera, but have yet to hear back on possible collaborations, discussion or testing validation.

Was a sessional instructor for CPSC453 - Introduction to Computer Graphics during Fall 2015, managing 3 TAs and 75 students. While an amazing opportunity and incredibly motivating and inspiring for computer graphics, it was very time consuming (attempting to teach well, organizing and creating lecture material, re-working existing materials, organizing and updating assignments, creating and grading exams).

Today: (And last couple of days) After winter break, have been re-visiting where I am in the code base, and what needs to be updated/corrected to improve results. Overall todos:

  • Improving visuals through KinFu integration - variant of Barnes/Perera's 2014 work
  • Supporting NaN values for feature points (ie. spatially or temporally occluded) - consider modern tracking methods?
  • Improve speeds (GPU, parallelization, use of transformations, dynamic/adaptive feature tracking based on success/structure?)
  • Joint network determination
  • Improved test cases

I've taken on a temporary part-time contract with a great software company downtown for the coming term. An artificial deadline for April has been set for evaluation of PhD progress for future directions.Read more...

January 06, 2016, at 06:38 AMerika

Friday, June 5, 2015

Yesterday: (Wrote it up yesterday).

Today: Added support for NaN values (ie. when feature points aren't always visible). Created test cases where feature points are visible 80% of the frames (5 frames)

7 Features 80% Vis
My 3books PCD Dataset

7 Features 100% Vis
My 3books PCD Dataset

SURF Only 100% Vis
My 3books PCD Dataset


June 05, 2015, at 02:49 PMerika

Thursday, June 4, 2015

Yesterday: Generated a few PCD examples, including one with moving antlers, 3 moving books and books "moving" in 3D. Had issues with insufficient feature points being detected.

Today: Incorporated support for blending multiple types of feature point into a single tracking for frames. Also added support to have the series keep keypoints that are occasionally hidden (based on percentage of frames, but could be otherwise). NaN values are inserted when they are not present. This can increase the number of keypoints by three-fold. Need to add support within rigid object segmentation for these NaN values. Also, it appears there are already point tracking algorithms, which I might want to look into. Moving antlers still aren't detected, but might with NaN values...Read more...

June 04, 2015, at 04:32 PMerika

Wednesday, June 3, 2015

Yesterday: (Last couple days) Struggled with getting PCL, boost, etc. to all work so that I can capture PCD files for use. Played with Ubuntu, Matlab, Windows set ups, with no easy results. Bugs with saving RGB colours on Matlab. Boost permissions issue with SD card storage on Ubuntu. Incorrect/corrupt dlls on Windows (XnCore, XnDDK).

Reference letter sent off. Sent off request for references from supervisors for CPSC453 instruction. Added a bit more math to adjacency matrix description within the report. Need to include visual explanations.

Today: Success on PCDs loading! (Ubuntu install). Will spend the rest of the day seeing how the feature points get extracted and 3D results. Submitted CPSC453 instructor application documents. Read more...

June 03, 2015, at 09:16 AMerika

Monday, June 1, 2015

Yesterday: (Last few days) Was able to get examples from Perera et al. to load, including visualizing their clustered results within my program. Able to load in pcd files and convert them to matrices for use by the program (saves only features that are always visible consistent with Perera et al's approach - ie. NaN values TBD). In turn, able to load this into my own program for clustering, etc. Good visual results, though low feature points extracted and only for few frames. Able to render point clouds and overlay feature points. Had an issue with the feature points overlay order (SortMethod depth vs childorder) in Matlab, which was resolved with a little guidance.

Attempting to get additional tests cases for better examples, but the helper functions built 2 years ago no longer work - updating libraries, etc. to get functional.

Submitted AITF report. Accepted for teaching CPSC453 conditional on formal application. Request for reference letter from undergraduate course. Request for formal report describing work to-date.

Today: Spent the last two days and today getting PCL and Boost libraries to work on Ubuntu with cmake. Finicky thing but finally seem to be able to get it compiling. Need to test that it will register the Kinect and attempt to generate new samples. Read more...

June 01, 2015, at 12:03 PMerika

Tuesday, May 26, 2015

Yesterday: Developed a conversion tool for pcd to matlab style matrices, including feature point detection (SIFT - could be other), matching, comparison against prior list with position updating or appending. Issue with handling duplicate matches. Will need to adjust system to support NaN vertices (ie. ones that are not visible either via kinect occlusion, or not visible in the scene); possibly tie in better with processing to use location estimates to search for subsequently matching features.

Today: Spent the whole day writing up final report for AITF scholarship. Was able to adapt from project proposal as well as scholarship application summary; good to have list tracking key accomplishments within a given year. While I'm not superwoman, I did accomplish things. Needing to get it written up though... Read more...

May 26, 2015, at 07:02 PMerika

Friday, May 22, 2015

Yesterday: (last couple of weeks) Adapted Perera et al.'s (2012) datasets for input into my own system. Included changes for thresholding. Checks on timing and accuracy, though not well validated yet. Will need to adjust to have solid accuracy checking. Broad overview of polynomial time algorithms for maximal cliques.

Good discussion to lay groundwork for multi-resolution processing.

3D Adjacency
Data from Perera et al.
(2012) 3books

2D View
Data from Perera et al.
(2012) 3books

Today: Wrapped up Perera et al. dataset adaption. After finding a bug in randomized ordering to ensure input order is inconsequential, I have success!!! See the animation (contains large amounts of "jitter" due to the different camera viewpoint in each frame) and adjacent images for object identification. Read more...

May 22, 2015, at 06:02 PMerika

Tuesday, May 12, 215

Yesterday: After Friday's struggle differentiating the maximal clique algorithm from my approximation approach, I determined that there doesn't appear to be much difference unless its application-specific. Have started implementing/adapting the maximal clique algorithm to speed up based on the conditions for 3D rigid object detection. Cursory results seem promising, though a wider selection of test cases needs to be explored.

Yesterday was spent getting Perera's implementation up and running (as it tends to). C-based implementation for maximal clique detection seems fast, but the datasets are small (and correspondences pre-determined). Requires further evaluation. Emailed Faramarz about discussion/direction.

Today: Start comparing and contrasting the maximal clique approach: re-arrange personal datasets for input into Perera's and similarly Perera's datasets into my own. Compare/time differences in maximal clique approaches/short-circuiting. Discuss/talk out with Faramarz how the approximate clique might be used for speedup. Read more...

May 12, 2015, at 06:08 AMerika

Friday, May 8, 2015

Yesterday: (last couple days) Continuing with paper writing - very slow progress. Started writing up the methodology. May need to re-evaluate how I'm using the approximate maximal cliques, and how best to justify it.

Today: Continue re-evaluating exactly how the approximate maximal cliques is being used, and justification for why its worthwhile in this situation. Write up how the distance matrix works, what it achieves, etc. Include the notes from Faramarz. Read more...

May 08, 2015, at 12:37 PMerika

BlogIt Side bar

Recently Written

erika: end hiding comments and categories

erika: end hiding comments and categories

Daily Scrum

erika: end hiding comments and categories

erika: end hiding comments and categories

Group-Specific Sidebar

Blix theme adapted by David Gilbert, powered by BlogIt