Skip to main content

Week 02

Nov 7​

CCTV​

  • will implement spatial correlation with frame with the same distance as counter distance
    • might change from same distance as counter distance to distance within 0.5m as counter distance etc in the future
  • will discard image with 1 m before counter distance (store 1 m as parameter) for the initial run
    • will change from 1m to other values in the future

Collimator​

  • BSM2 in Matlab

BSM2 in Matlab

  • BSM2 in Collimator

BSM2 in collimator

  • I am still working on ADM block inside anaerobic digester block.
  • Note that if you are reading parameter from csv file using python script block, if the value is an int, you will need to convert it to a float in order to assign it to the output port. In addition, the python script block are able to read csv files with 100 columns.

Nov 8​

  • email to Joakim
  • email to Elsevier people
  • write read about intro to linux
  • write readme about intro to computer hardware
  • write readme how to use compute canada (compute canada for dummies)
  • Talk to deven to get some idea creating the bibilography in latex?
  • Create label file for first 3 videos obtained from Azurze
  • Submit a job to compute canada using the first 3 videos and label obtained from google and azurze

Nov 9​

  • I submitted two runs to compute canada
    • Job 29: train and validate with images from video 1, 12, 14, 20 of the Lynn Valley Oct 6 2021 folder
    • Job 30: train and validate with images from all 33 videos of the Lynn Valley Oct 6 2021 folder
  • I finished building Sh2 solver block
    • Instead of doing ODE, they used Newton Raphson to solve for hydrogen solution (Sh2). They used the same approach for pHsolver block.
    • Collimator has discrete integrator block; however, it uses Euler method which only require the function as an input.
    • Newton Raphson needs both function and derivative of the function.
    • I will need to compare the values to Matlab output once I finish building these two blocks.

Nov 10​

  • Email Dr. Lence about attending the Metro meeting for a few minute

    • Show the streamlit app (Sewer ML data) (backup Deven and Pavan)

  • Streamlit with the new model (Ask Deven how to add the new model)

  • Follow up with Deven about unix and computer hardware for dummy

  • right now, Pacp are using access database as exchange data standard. Is there going to be other format in the future, sqlite

  • Introduction:

    • Internship with GQC based in cincinnati, ohio
    • Researching on AI application that would assist in inspection during quality control? The point is not to replace human and help identify the defect that might be missed during the inspection.
    • help with the inspection
    • If anyone has video that did not have conflict with client that and interesting in running with our model, please let me know

  • Send the create image label code for Pavan to review

  • Make sure the code follows the same logic as excalidraw.

  • need to merge access database with video footage because not all defect was captured in the video annotation.

    • Export PACP_Conditions
    • Export PACP_Inspections
    • Export PACP_Media_Inspections
    • Merge the info to each label file using video_name, distance where defect is found.
    • Instead of just using counter distance, you would also use the distance found in the access database

  • Collimator: There is a problem with reading csv file. Have asked Brain to fix this.

Last updated on by Viraj