Phenology GUI

This GUI is for displaying images and selecting region of interest (ROI).

requirement

Python 3 and a few external modules (install them with pip install <module> if Python complains).

Instruction

1. Place utils.py and session.py in the same directory and run session.py.
2. You have the option to load an output csv file from previous session or start a new session.
   • The output file from previous session should not have been altered! Work with copies if altering the file is necessary
   • If starting a new session, select all image files by using Ctrl or Shift buttons.
3. Marking ROI
   • Select ROI using the radio option.
   • Create a polygon on the image.
     • Left click adds a point
     • Right click finalizes the polygon
   • Repeat above steps. You can reset the ROIs with “Clear ROI” button.
   • When you’re done with the image, click “Finalize Image”. This will save the ROI for the next image (“Apply Previous ROI” button).
4. Output
   • Click “Save and close” to save the output csv file.
   • The output will be a csv file with the iamge metadata, ROI coordinates, and ROI statistics
     • Note: ROI coordinates will be a list of (x,y) values
       • i.e. [(3,4), (10, 10), (5,2)]
   • This is the file you can load a session with.
5. Hotkeys
   • Enter: finalize image
   • Space: apply previous ROI
   • Right arrow: skip to next image

Notes

Modifying ROI types

To modify the ROI types, you can edit the ROI_TYPES and ROI_COLORS variables in utils.py.

Modifying the ROI statistics

This is more complicated. You have to modify the STATS variable in utils.py, but the output file save the result. I need to test this.

Modifying image canvas size

This might be useful if a large image is causing issues with the size of your app. You can change the size of region dedicated for displaying the image (default is 1600x1200). This is in utils.py under CANVAS_SIZE variable.

Issues

Get Camera_id and image Date

RESOLVED Camera_ID

Assume that the input directory is the camera_id.

RGB for entire image

IO - single input and output per camera

The input will be a folder associated with a camera.

Read all of .jpg files in the subdirectories

RESOLVED Large memory requirement

The current implementation keeps all the images around. i.e. for 200 images, I had 4gb memory overhead.

Fix this so that only current image is opened.

Arrange the images in the folder by date

Output: create indicator for change in camera position

Create a button to save without closing

Don’t close app on “save”.

Prompt for a new folder

Add scrollbar to the canvas

Image scrollbar is on the top and left. When the image is too large you can’t access the right and bottom side of the app.

ISSUE I need access to the NAS for the actual data

RESOLVED Copied some data but still need access

Test modifying ROI statistics

ISSUE Prev_ROI Function

If image is finalized without ROI, it overwrites the previous ROI. This is a safe way but it might be inconvenient.