v0.3.1

Some fixes:

Fix Cellpose v3 compatibility: update API calls and enforce v3 requir…

Fix: Check if tmidas-env exists before creating to prevent installati…

Fix binary compatibility issues by installing numpy/scikit-image/tiff…

Add flow_threshold and cellprob_threshold parameters to Cellpose inst…

Fix default value handling in user_welcome.py

Fix CPU/GPU switching in cupy-dependent scripts

v0.3.0 - GPU Error Handling & CPU Fallback

What's New

• GPU error handling: All pipelines now gracefully fall back to CPU when GPU/drivers are unavailable
• User notifications: Welcome screen detects GPU availability and warns users about CPU-only mode
• No more crashes: nvidia-smi failures and CUDA errors are caught and handled transparently

Updated Pipelines

• SAM segmentation (2D, NDPI cropping)
• Instance segmentation
• Cell tracking (TrackAstra)

All torch-based scripts now automatically use CPU if GPU initialization fails.

Fixes

• Fixed crashes on systems without GPUs or with driver issues
• Protected torch.cuda calls throughout the codebase

production-ready release v0.2.0

Full Changelog: v0.1.9-beta...v0.2.0

v0.1.9-beta

updated user_welcome