[BUG] Segmentation Fault when using the library to remove background form an image
umutyazgan opened this issue · 2 comments
umutyazgan commented
Describe the bug
I get a "Segmentation fault" when I try to remove the background from a PIL image, using the library with GPU support.
To Reproduce
Steps to reproduce the behavior:
- Write this script:
from rembg import remove
from PIL import Image
input_path = 'input.png'
output_path = 'output.png'
input = Image.open(input_path)
output = remove(input)
output.save(output_path)
-
Use this input image:
-
run it:
python rembg_test.py
OMP: Info #276: omp_set_nested routine deprecated, please use omp_set_max_active_levels instead.
2024-01-24 16:35:14.315665635 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:14 WARNING] onnx2trt_utils.cpp:364: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
2024-01-24 16:35:16.588285820 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:16 WARNING] TensorRT was linked against cuBLAS/cuBLAS LT 11.5.1 but loaded cuBLAS/cuBLAS LT 11.3.0
2024-01-24 16:35:17.065906556 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:17 WARNING] TensorRT was linked against cuDNN 8.2.1 but loaded cuDNN 8.0.5
2024-01-24 16:35:17.069822559 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:17 WARNING] Detected invalid timing cache, setup a local cache instead
2024-01-24 16:35:50.336724380 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:50 WARNING] TensorRT was linked against cuBLAS/cuBLAS LT 11.5.1 but loaded cuBLAS/cuBLAS LT 11.3.0
2024-01-24 16:35:50.339813076 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:50 WARNING] TensorRT was linked against cuDNN 8.2.1 but loaded cuDNN 8.0.5
2024-01-24 16:35:50.351736494 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:50 WARNING] TensorRT was linked against cuBLAS/cuBLAS LT 11.5.1 but loaded cuBLAS/cuBLAS LT 11.3.0
2024-01-24 16:35:50.355949247 [W:onnxruntime:Default, tensorrt_execution_provider.h:77 log] [2024-01-24 16:35:50 WARNING] TensorRT was linked against cuDNN 8.2.1 but loaded cuDNN 8.0.5
Segmentation fault
Expected behavior
Shoud run without a SegFault and remove the background from the image.
Images
Another input image I tried:
OS Version:
Debian GNU/Linux 11 (bullseye) x86_64
Kernel: 5.10.0-27-cloud-amd64
Rembg version:
v2.0.53
Additional context
The machine is on Google Compute Engine.
GPU: NVIDIA Tesla T4 16 GB
nvidia-smi
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 545.23.08 Driver Version: 545.23.08 CUDA Version: 12.3 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Tesla T4 On | 00000000:00:04.0 Off | 0 |
| N/A 37C P8 12W / 70W | 2MiB / 15360MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+
Some packages in my conda environment that may be relevant:
cuda-cudart 12.1.105 0 nvidia
cuda-cupti 12.1.105 0 nvidia
cuda-libraries 12.1.0 0 nvidia
cuda-nvrtc 12.1.105 0 nvidia
cuda-nvtx 12.1.105 0 nvidia
cuda-opencl 12.3.101 0 nvidia
cuda-runtime 12.1.0 0 nvidia
cudatoolkit 11.1.1 hb139c0e_12 conda-forge
imageio 2.9.0 pypi_0 pypi
imageio-ffmpeg 0.4.4 pypi_0 pypi
nvidia-cublas-cu12 12.1.3.1 pypi_0 pypi
nvidia-cuda-cupti-cu12 12.1.105 pypi_0 pypi
nvidia-cuda-nvrtc-cu12 12.1.105 pypi_0 pypi
nvidia-cuda-runtime-cu12 12.1.105 pypi_0 pypi
nvidia-cudnn-cu12 8.9.2.26 pypi_0 pypi
nvidia-cufft-cu12 11.0.2.54 pypi_0 pypi
nvidia-curand-cu12 10.3.2.106 pypi_0 pypi
nvidia-cusolver-cu12 11.4.5.107 pypi_0 pypi
nvidia-cusparse-cu12 12.1.0.106 pypi_0 pypi
nvidia-nccl-cu12 2.18.1 pypi_0 pypi
nvidia-nvjitlink-cu12 12.3.101 pypi_0 pypi
nvidia-nvtx-cu12 12.1.105 pypi_0 pypi
onnxruntime 1.16.3 pypi_0 pypi
onnxruntime-gpu 1.16.3 pypi_0 pypi
opencv-contrib-python 4.9.0.80 pypi_0 pypi
opencv-python 4.9.0.80 pypi_0 pypi
opencv-python-headless 4.9.0.80 pypi_0 pypi
pillow 8.1.0 pypi_0 pypi
python 3.8.5 h1103e12_9_cpython conda-forge
pytorch 2.1.2 py3.8_cuda12.1_cudnn8.9.2_0 pytorch
pytorch-cuda 12.1 ha16c6d3_5 pytorch
torch 2.1.2 pypi_0 pypi
torchaudio 2.1.2 py38_cu121 pytorch
torchdiffeq 0.2.3 pypi_0 pypi
torchmetrics 1.3.0.post0 pypi_0 pypi
torchsde 0.2.6 pypi_0 pypi
torchtriton 2.1.0 py38 pytorch
torchvision 0.10.1 py38_cu111 pytorch