Open zsylabtip1 opened 1 year ago
我在运行extract_features.py时,产生了错误,下面是完整的错误信息: 请问有什么办法解决这个问题吗?
/root/anaconda3/envs/yiz_py/bin/python3.8 /data/yiz/FasterRcnn/extract_features.py Traceback (most recent call last): File "/data/yiz/FasterRcnn/extract_features.py", line 17, in <module> from utils.extract_d2features import extract_feat_d2_start File "/data/yiz/FasterRcnn/utils/__init__.py", line 1, in <module> from .utils import save_features File "/data/yiz/FasterRcnn/utils/utils.py", line 7, in <module> from bua.caffe.modeling.layers.nms import nms File "/data/yiz/FasterRcnn/bua/__init__.py", line 1, in <module> from .d2 import add_attribute_config File "/data/yiz/FasterRcnn/bua/d2/__init__.py", line 5, in <module> from .modeling.roi_heads import AttributeRes5ROIHeads File "/data/yiz/FasterRcnn/bua/d2/modeling/roi_heads.py", line 20, in <module> from bua.caffe.modeling.box_regression import BUABox2BoxTransform File "/data/yiz/FasterRcnn/bua/caffe/__init__.py", line 2, in <module> from .modeling.backbone import build_bua_resnet_backbone File "/data/yiz/FasterRcnn/bua/caffe/modeling/__init__.py", line 3, in <module> from .roi_heads import BUACaffeRes5ROIHeads File "/data/yiz/FasterRcnn/bua/caffe/modeling/roi_heads.py", line 17, in <module> from .fast_rcnn import BUACaffeFastRCNNOutputs, BUACaffeFastRCNNOutputLayers, BUADetection2FastRCNNOutputs, BUADetectron2FastRCNNOutputLayers File "/data/yiz/FasterRcnn/bua/caffe/modeling/fast_rcnn.py", line 15, in <module> from .layers.nms import batched_nms File "/data/yiz/FasterRcnn/bua/caffe/modeling/layers/nms.py", line 3, in <module> from bua.caffe.modeling import _C ImportError: cannot import name '_C' from partially initialized module 'bua.caffe.modeling' (most likely due to a circular import) (/data/yiz/FasterRcnn/bua/caffe/modeling/__init__.py) Process finished with exit code 1
Your torch version or cuda version maybe wrong. If you use RTX30xx or higher, It may fail to build. I recommand you to use cuda 10.1. Here is my solution, you can try it: torch1.5+cu 101+cp36 on 2080ti, apex version--22.04, detectron2--v0.3, maybe you could try this.
这是来自邮箱的假期自动回复邮件。您好,我最近正在休假中,无法亲自回复您的邮件。我将在假期结束后,尽快给您回复。
我在运行extract_features.py时,产生了错误,下面是完整的错误信息: 请问有什么办法解决这个问题吗?