iris
iris copied to clipboard
Semi-automatic tool for manual segmentation of multi-spectral and geo-spatial imagery.
Currently, when multiple users annotate images at the same time, different users get the same images. I believe this is undesirable behavior. Re-annotation should only occur after all images have...
Document experience upgrading to `lightgbm==4.1.0` which is required for Mac wheels - Get `ModuleNotFoundError: No module named 'sklearn'` - add to requirements - `AttributeError: module 'hashlib' has no attribute 'scrypt'`...
I needed to update permissions: ``` ~/iris chmod 777 /teamspace/studios/this_studio/iris/demo ``` Then run: ``` iris docker run -p 5000:5000 -v /teamspace/studios/this_studio/iris/demo:/demo/ --rm -it iris demo ``` However the app is...
I want to annotate change detection and generate 8 bit`mask.png` which is single channel and 0 is `no-change` and 255 is `change`, which I think is pretty standard format. However...
Config would require the index of the bands to used from the input. Ref https://torchgeo.readthedocs.io/en/stable/tutorials/indices.html
Currently the zoom level is hard coded [I think here](https://github.com/ESA-PhiLab/iris/blob/b5e07329ecf8d91b3eb4b3d4f05bf5229f657557/iris/static/javascripts/views.js#L450) but it would be nice if this could be configured like: ```yaml "Bing": { "description": "Aerial Imagery", "type": "bingmap", "zoom_level":...
As option in config, to not see this:
On updating my config and restarting: ``` "GET /segmentation/load_mask/SE320572405 HTTP/1.1" 404 ``` Any tips to debug?
Running `iris demo` I annotate each class and hit the AI button: I get: ``` Fit options: {'mask_encoding': 'rgb', 'score': 'f1', 'prioritise_unmarked_images': True, 'unverified_threshold': 1, 'test_images': None, 'ai_model': {'bands': None,...
This would allow displaying the annotations on a clear or white background, which I think may be easier to visualise than when overlaid on an image. Not currently an option