Yesterday: Re-reviewed and provided suggestions for improving conciseness of the PYXIS CRD application. Spent some time looking at graphics announcements and events taking place in Calgary, specifically surrounding GIRAF Fest.

Correctly re-ordered transformation application, mixed up points for accurate randomization, and implemented distance-based pruning. Example of two spheres with 15 points apiece being segmented (video). Pausing is for animation, and not reflective of the seemingly online runtime.

Today: Visualize graph representation of these point clouds. Implement transformation-based pruning. Generate a file-loading for test cases so there is consistency in process validation.

Looking into getting the Kinect running within Matlab (link) and get back to Matlab group about the various toolboxes. Read more...

Yesterday: (Wednesday) Reviewed PYXIS CRD to assist with group grant applications. Brief update meeting with Faramarz. Will need to show off results once I have them. May need to figure out how to perform graph representation within Matlab (perhaps there are existing tools) or parallelization for transformation calculations.

Have basic rotation of points implemented. Need to it to rotate about the object's midpoint and validate other rotations (eg. 0^o).

Today: Have 3D rotations for rigid body motions about object's center. Mix up the 3D points, and be able to identify the points which transform rigidly with respect to each other. Visualize the process.

Review the PYXIS CRD for ways to reduce the page count - ie. succinct and precise wording.

Looking into getting the Kinect running within Matlab (link) and get back to Matlab group about the various toolboxes. Read more...

Yesterday: Group meeting on simple sketching for 3D building modeling. Adjusted the matlab code so it views the rigidly transforming point clouds better by default. Installed Matlab on the laptop and connected with Matlab about working with their Kinect and CV libraries.

Took a look at a couple of wilderness cameras. There seems to be night vision and daytime captures. Unsure how to build custom-configured (heat, night, daytime, depth) and be able to define an SDK for self-use, though the price point would be steep ($6k for a drone, $300-800 for GoPro, $300-400 for wilderness camera).

Today: Work on incorporating interpolation of 3D rotations for rigid body motions. Mix up the 3D points, and be able to identify the points which transform rigidly with respect to each other. Visualize the process.

Looking into getting the Kinect running within Matlab (link). Read more...

Yesterday: Added local animation groups to cg_yyc. Unable to track some individuals such as Cam Christiansen of Anlanda. May need to keep an eye on the Banff Center as well. Conversation with the supervisor about long-term future directions.

Today: Get 3D plotting of stationary feature points within Matlab. Then work on incorporating 3D transformations. Place hold in library for UI/UX books or find one online for continued motivation. Look into purchasing wilderness cameras. Read more...

Yesterday: Added a couple extra companies to cg_yyc. Have been transferring the logic behind rigid segmentation over to Matlab as suggested by the supervisor. At this point have vertices of obj files and spheres being loaded, and translating across the screen.

Today: Get interpolation of axis-based rotation functioning for synthetic Matlab examples (ie. SLERP?). Randomize, but have mapping for generated points for more sporadic synthetic layout for rigid segmentation. Start applying segmentation stages. Read more...

Yesterday: Discovered lists on Twitter. Have constructed one for general computer graphics. Have created a new user name cg_yyc to attempt to track computer graphics news in and around Calgary. A list of companies (though missing Halliburton's Pinacle and the Computer Modelling Group) and affiliates within the computer vision realm. There really should be a yyc_siggraph feed, but then I'd have to start a local chapter.

Starting to read Smashing UX Design: Foundations for Designing Online User Experiences by Jesmond Allen and James Chudley, 2012. Its available online through the University Library.

At the end of the day, tweaking the manually generated list of points from Excel, figuring out how load the custom file into Matlab, and eventually display the points, I discovered that generating N random 3D points on a sphere has already been solved. Now to incorporate the proof-of-concept for rigid body detection.

Today: Get 3D plotting of stationary feature points within Matlab. Then work on incorporating 3D transformations. Place hold in library for UI/UX books or find one online for continued motivation. Read more...

Yesterday: Filed away some papers that were of interest, tracking them accordingly within the notes documents. Met with Faramarz, and discussed potential break or alternative arrangement for research. Discussion included suggestion to work with Matlab to get code up and running faster. Manually created sample feature points on a sphere and am working on loading them into Matlab. Also manually generated simple descriptive file for 3D transformations (rotation based on a user-provided unit vector, akin to quaternions).

Today: Get 3D plotting of stationary feature points within Matlab. Then work on incorporating 3D transformations. Place hold in library for UI/UX books or find one online for continued motivation. Read more...

Yesterday: Motivational challenges and indecision on which approach to take for creating synthetic 3D examples has limited progress. Choices include Qt/Linux dev environment, or potentially exploiting Unity. Either way I need to "just" generate a list of 3D feature points, and then decide on valid transformations which prevent overlap. Ideally this would support simple .obj files. Motivational issues may require looking into UX/UI textbooks and time browsing Slashdot and 3D graphics/reconstruction sites.

Today: Get a simple collection of 3D feature points generated, even if manually done. Read more...

Yesterday: Collected materials from previous years' conference organization and passed them along to the colleague who is more directly associated with the work, and will be taking over this responsibility. Nominal work done on the synthetic 3D examples.

Today: Continue developing synthetic 3D examples to work with. Both for proof-of-concept of transformation extraction, and also for later RGB-DN development. Read more...

Yesterday: Wrote out high level requirements for the various data sets required for 3D articulated rigid body validation. Started laying out codebase for generating said examples. RGB-DN examples maybe trickier to synthetically generate (or to work with; and the 2D images aren't necessarily available when a stop-motion style comprehensive scan is generated). Temporarily given up getting PCL and OpenCV working on any machine.

Today: Continue developing synthetic 3D examples to work with. Both for proof-of-concept of transformation extraction, and also for later RGB-DN development. Read more...