Created by: SangerY

win7+cuda10+cudnn7.6，模型为faster_rcnn_r101_vd_fpn_1x，导出模型预测，执行中断也不报错。 使用如下代码导出模型： python tools/export_model.py -c configs/faster_rcnn/faster_rcnn_r101_vd_fpn_1x.yml
--output_dir=./inference_model
-o weights=output/faster_rcnn_r101_vd_fpn_1x/best_model 再使用python deploy/python/infer.py测试视频，结果如下，也不报错，就自动退了。 Running Arguments image_file: model_dir: models/ output_dir: I:/results/ run_benchmark: False run_mode: fluid threshold: 0.3 use_gpu: True video_file: I:/0001.mp4

Model Configuration Model Arch: RCNN Use Padddle Executor: False Transform Order: --transform op: Normalize --transform op: Resize --transform op: Permute --transform op: PadStride

detect frame:1

将infer_cfg.yml中的use_python_inference改为true后，也是直接跳出来了，不报错。 Running Arguments image_file: model_dir: models/ output_dir: I:/results/ run_benchmark: False run_mode: fluid threshold: 0.3 use_gpu: True video_file: I:/0001.mp4

Model Configuration Model Arch: RCNN Use Padddle Executor: True Transform Order: --transform op: Normalize --transform op: Resize --transform op: Permute --transform op: PadStride

W0721 14:08:58.003999 19100 device_context.cc:252] Please NOTE: device: 0, CUDA Capability: 61, Driver API Version: 10.0, Runtime API Version: 10.0 W0721 14:08:58.016000 19100 device_context.cc:260] device: 0, cuDNN Version: 7.6.