developer0hye / Yolo_Label

GUI for marking bounded boxes of objects in images for training neural network YOLO
MIT License
509 stars 114 forks source link

Suggestions that will improve the app significantly #64

Open FatHyZayed opened 8 months ago

FatHyZayed commented 8 months ago
  1. Add Zoom in/out feature
  2. Key binding feature to bind let's say Class number 1 with keyboard H button and the Visualize button to another button like Mouse MB4.
  3. Size Adjust of the picture with the right panel that has the Class names and colors because some class names are too long and can't adjust that.
  4. a Timer feature that counts how many hours you've been using the program, it starts counting only when the program is focused and you can reset it as you like.
  5. an Undo button would also be helpful, Ctrl + Z, as if you removed something by mistake.

i hope these are add it because this is the best program i've seen that does the job pretty well.

ForrestFire0 commented 4 weeks ago

another idea - option to duplicate the previous annotations from the previous image to the next.

In addition to that, a simple open cv feature tracking add on that will look at the image for a transform to the next, if it sees one, move the last boxes by that transform. I am labelling images that are sampled at 1 hz from video, so a lot of my boxes are very close to the location of the last box.

(I know that takes away from the dirt simple nature of the tool that makes it so great)

Wish this was written in python! Then I could do it myself. I don't wanna muck about with opencv in cpp though :\