Comments (5)
For anyone who encounters this issue, the current workaround is:
pip install torchgeo lightly!=1.4.26
It looks like this is going to be fixed in the next lightly release: lightly-ai/lightly#1487
from torchgeo.
Thanks for the bug report! We also noticed this when dependabot tried to update our lightly dependency in #1823. Here is a rundown of how this bug surfaced. It's actually pretty complicated (and interesting!)
The issue started when lightly v1.4.26 added an optional dependency on timm in lightly-ai/lightly#1479. If timm is available, lightly exposes additional parts of the API and imports certain features of timm. However, the features it uses are only available starting in timm v0.9.9: huggingface/pytorch-image-models#1996.
At the same time, TorchGeo depends on segmentation-models-pytorch, which requires timm v0.9.2: https://github.com/qubvel/segmentation_models.pytorch/blob/v0.3.3/requirements.txt#L4. We've asked them many times not to pin to a specific version of timm, but to no avail. The library itself may not even be maintained anymore, there haven't been any updates in quite some time.
So the short-term fix is to pin TorchGeo to depend on lightly<=1.4.25
. This will be included in the next TorchGeo patch release, likely some time in the next week. A medium-term fix would be for lightly to first check the version of timm available and only try to use it if it's timm v0.9.9+. I'm not sure if they will be interested in the additional code complexity to check this, but I'll suggest it anyway. A long-term fix would be to somehow resurrect or replace segmentation-models-pytorch so we can finally use modern versions of timm. I'll also explore this, but I won't get my hopes up.
from torchgeo.
This is now fixed in main and will be included in the next release. It's also been fixed in lightly itself (lightly-ai/lightly#1487), so future lightly releases will be fine. We're still considering migrating from smp to torchseg to remove the timm version constraint, which will also alleviate this issue.
from torchgeo.
from torchgeo.
FYI, lightly 1.5.0 has now been released and fixes this bug.
from torchgeo.
Related Issues (20)
- Incompatible image size with RandomGeoSampler HOT 3
- Easier way to use Data Processing steps outside of datamodule HOT 4
- Benchmarking of all pre-trained weights HOT 4
- Add instructions on downloading the DeepGlobeLandCover dataset HOT 5
- SSL Weight Decay HOT 6
- Migrate from Radiant MLHub to Source Cooperative HOT 13
- Datamodule augmentation defaults HOT 8
- NCCM checksum error HOT 6
- Support additional SatlasPretrain models. HOT 6
- Document significance of macro vs micro averaging HOT 3
- Add BalancedRandomGeoSampler balancing positives and negatives HOT 2
- Add support for Lightning Streaming Dataset HOT 14
- OSCDDataModule initialises with batch_size 1, ignoring the configured batch_size HOT 4
- Add `ignore_index` support for Jaccard Loss HOT 1
- Unpin torch, use a min or range? HOT 4
- trainers.segmentation JaccardLoss receiving num_classes, should be a List[int]? HOT 8
- GeoDataset: non-deterministic behavior HOT 5
- Sentinel 2 dataset can't see files downloaded from Copernicus Browser - filename doesn't fit regex HOT 1
- Errors & improvements in Metrics descriptions HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from torchgeo.