RuntimeError: indices should be on the same device as the indexed tensor in MMRotate tutorial

[ramondalmau]

Description:
I encountered a RuntimeError when following the MMRotate tutorial. The error occurs when running inference_detector(model, img). My model is on cuda:0, but it seems like some indices are on the CPU, causing a mismatch.

Error Message:
RuntimeError: indices should be either on cpu or on the same device as the indexed tensor (cpu)

Environment:

• CUDA: 12.4
• PyTorch: 2.6.0
• MMDetection: 2.28.2
• MMCV: 1.7.2
• MMRotate: 0.3.4
• GPU: NVIDIA RTX A4500
• OS: Red Hat Enterprise Linux release 8.9 (Ootpa)

Steps to Reproduce:

1. Install MMRotate following the official guide.
2. Load a model and set it to cuda:0.
3. Run inference_detector(model, img).
4. The error occurs.

Expected Behavior:
The inference should run without device mismatch issues.

What I Tried:

• Reinstalling dependencies.

Let me know if you need additional logs or debugging outputs.