A new object detection model is coming...

[hazegreleases]

Hello, i wanted to talk about the new model that i am working on and the integration to JIETStudio and the current "stall" of updates to the app.

so for the last 2 months, the updates to the JIETStudio app has been stopped, this is temporary as my computer is not in a state of being "useful", because it trains the cv model i am working on. And the cv model building has took a longer time than what i have expected because i had to retrain from scratch change the architecture train change train change untill "perfection", and i believe i have "perfected" what i did. In probably 8-9 days the training should be finished.
Just as a heads up; the model will be a 3.4M parameter, NMS-FREE model that will be a yolo drop-in style model, it does NOT rely on any ultralytics in any way but is just inspired from ultralytics newer models (specifically yolo10-11 and 26).

So what happens to JIETStudio?
After the training finishes on my computer, i will add more QoL features to JIETStudio which are currently absent. And in a few weeks to a month, i will be focusing on polishing and making my own model easier to use faster to inference and better at detection which then i will push the project to github and will put it to be easily downloadable trough pip,which then the model will be integrated on-to JIETStudio to be in use.

thanks for anybody who uses JIETStudio and sorry for any typos that are in this paragraph as im typing this trough my phone as my pc is rendered useless to anything currently AKA training the model.

And if you have any feedback or "must have features" for the JIETStudio, i would love to implement them as many as possible!

hazegreleases out.