v 1.6

• Add Cellpose 0-2mm model using PyTorch + SAM2 (Meta's Segment Anything Model 2)
• Model weights not included due to file size (1.19GB exceeds GitHub limit)
• Place cellpose_model_weights and config.json into models/oyster_cellpose_0-2mm/ manually
• See README for instructions
• Add StarDist 0-2mm model for CPU users
• All 4 built-in models now load automatically – no file browsing required
• Add full NVIDIA CUDA GPU support with CPU/GPU switching via dropdown
• Add live GUI progress bar during prediction
• Add terminal progress bar (tqdm) for developer monitoring
• Fix GPU/CPU model cache bug – switching devices no longer loads stale model
• Model settings and parameters now persist when switching between images
• CPU thread optimization using all available cores
• Update README with model selection guide, GPU setup, Cellpose weights instructions

Windows Download

The Windows executable is too large for GitHub.
Download here: https://drive.google.com/file/d/1iP9mH5el4KZpSA7uq3ees4OkaIzo2M7o/view?usp=sharing

• Place cellpose_model_weights and config.json into models/oyster_cellpose_0-2mm/ manually

v 1.5

Final release for version 1 of app, further releases will be included version 2

v1.4

Preview for the upcoming release. This includes a lot of changes for the GUI and also a binary distribution for Raspberry Pi's

What's Changed

• GUI Changes and new/updated features including camera support by @dhdschool and @MaxShroyer
• Distribution and setup guide for raspberry pi by @MaxShroyer

Full Changelog: v1.3...v1.4.0

LSU Math Clinic GUI v1.3

• Select Model now available by dropdown, manual selection available inside dropdown selection
• The new tab "File Names" in Excel Window, contains file and folder naming for export
• The new tab "File Save" in the Settings Window, contains save settings for runtime and enables files, folders, and models to load on startup
• The new tab "Program" in the Settings Window, contains version updates and user guide buttons
• The app is now packaged within a folder named app, which includes the exe and a folder named _internal.
• _internal folder enables save settings to work. The exe file must run in the same folder as _include or the program will not work.

Devision GUI v1.2

• Switched from CSV to Excel
• Added settings menu
• Added functionality for classification models developed by the Devision team
• GPU Support! In order to utilize the GPU, you need a CUDA compatible GPU by Nvidia listed here under CUDA-Enabled GeForce and TITAN Products. The left column is for desktop GPU's and the other column is for laptop GPU's.
• Instructions: Download the .exe and run it on your computer. The output folder will appear in the directory where the .exe is located.
• If you have any problems, please submit then under the Issues tab in this GitHub repository

Devision GUI v1.1

Rearranged buttons for ease of use.
Added functionality for csv output.
Added a button to clear currently selected images.
Changed output image to now display color.
Changed default model to none.
Added help button with information about buttons.

Devision GUI v1.0

Initial release of our GUI.