Week 12 - Adding options to interface

2 minute read

Adding more options

I have been improving the GUI Interface this weeks and integrating new options to the user. All the code can be found in Github

Using PySimpleGUI

I added some extra options for future uses (Binary, HUE, Enhance). Those can be loaded in live with a slider, so for future uses the user can be more precise with image preprocessing, regarding the background and colors in the scene.

NEW OPTIONS ADDED:

  • FPS.
  • Frame by frame pressing key ‘n’ predictions with recorded videos.
  • Binary, HUE and Enhance for Live mode previous to analysis.
  • Predefined configuration to Preproccesing the scene which consists in HSV -> GRAY -> BINARY -> EROTION -> dilation_image
  • Use the predefined model and get predictions (Not getting the correct values now)
  • Create Logs with information about the predictions. BUFFER, REAL VAlUES and PREDICTED VALUES
Cap0
Cap1
Cap2
Binary
HUE
Enhance

The logs files have the next information (example below): LOG - Mon Nov 29 15/02/54 2021.log

BUFFER

[[37 0] [37 1] [36 2] [36 3] [35 3] [35 4] [34 6] [34 7] [34 9] [33 11] [33 13] [32 15] [32 17] [31 18] [31 20] [30 22] [30 24] [29 26] [29 27] [28 29] [28 31] [27 33] [27 35] [27 36] [26 38] [26 40] [25 42] [25 44] [24 45] [24 47] [23 49] [23 51] [23 52] [22 54] [22 56] [21 57] [21 59] [21 61] [20 62] [20 64] [19 66] [19 67] [18 69] [18 71] [17 72] [17 74] [17 76] [16 77] [16 79] [15 80]]

REAL VALUES

[[ 15 82] [ 15 84] [ 14 85] [ 13 87] [ 13 88] [ 13 90] [ 12 91] [ 12 93] [ 12 95] [ 11 96] [ 11 98] [ 10 99] [ 10 101] [ 10 102] [ 9 104] [ 9 105] [ 9 107] [ 8 108] [ 8 110]]

PREDICTED VALUES

[[ 28 99] [ 23 105] [ 4 87] [ -1 92] [ 4 100] [ 6 103] [ 7 104] [ 8 105] [ 8 106] [ 9 106] [ 9 107] [ 10 107] [ 11 108] [ 12 108] [ 12 108] [ 6 101] [ 8 102] [ 9 102] [ 9 101]]

Again the predictions are not working propertly so I need to improve our generated images used to train other Network. We are focusing in the GUI experience and when everything looks clean, organized and work properly, it’s time to focus on predictions. Next steps will be continue with GUI and come back to generate new datasets to get better models and better predictions finally.