运行报错 find_blob_index_by_name images failed

[wangyuou]

您好@nihui

我使用yolov5生成一个best.pt,然后使用命令python export.py --weights best.pt --int8 --include onnx --simplify导出onnx模型，再通过pnnx best.onnx inputshape=[1,3,640,640]生成了best.ncnn.bin和best.ncnn.param文件，替换这个工程的文件，点击识别没任何反应日志输出如下： 35.89ms detect 2024-10-10 18:20:29.261 6891-6905 OpenGLRenderer com.tencent.yolov5ncnn I Davey! duration=9223260857263ms; Flags=0, FrameTimelineVsyncId=392958, IntendedVsync=111179591303092, Vsync=111179641303090, InputEventId=801281897, HandleInputStart=111179645622482, AnimationStart=111179645675451, PerformTraversalsStart=111179646356805, DrawStart=111179646678836, FrameDeadline=111179607969758, FrameInterval=111179645602065, FrameStartTime=16666666, SyncQueued=111179648169409, SyncStart=111179648266076, IssueDrawCommandsStart=111179654145555, SwapBuffers=111179655800868, FrameCompleted=9223372036854775807, DequeueBufferDuration=30677, QueueBufferDuration=729323, GpuCompleted=9223372036854775807, SwapBuffersCompleted=111179657125034, DisplayPresentTime=0, 2024-10-10 18:20:42.280 6891-6891 ncnn com.tencent.yolov5ncnn W find_blob_index_by_name images failed 2024-10-10 18:20:42.280 6891-6891 ncnn com.tencent.yolov5ncnn W Try 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W ex.input("in0", in0); 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W find_blob_index_by_name output failed 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W Try 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W ex.extract("out0", out0); 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W find_blob_index_by_name 781 failed 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W Try 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W ex.extract("out0", out0); 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W find_blob_index_by_name 801 failed 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W Try 2024-10-10 18:20:42.281 6891-6891 ncnn com.tencent.yolov5ncnn W ex.extract("out0", out0);

[wangyuou]

PS E:\yolov5> pnnx best.onnx inputshape=[1,3,640,640] pnnxparam = best.pnnx.param pnnxbin = best.pnnx.bin pnnxpy = best_pnnx.py pnnxonnx = best.pnnx.onnx ncnnparam = best.ncnn.param ncnnbin = best.ncnn.bin ncnnpy = best_ncnn.py fp16 = 1 optlevel = 2 device = cpu inputshape = [1,3,640,640]f32 inputshape2 = customop = moduleop = ############# pass_level0 onnx inline_containers ... 0.00ms eliminate_noop ... 0.17ms fold_constants ... 0.14ms canonicalize ... 0.24ms shape_inference ... 242.35ms fold_constants_dynamic_shape ... 0.11ms inline_if_graph ... 0.03ms fuse_constant_as_attribute ... 0.18ms eliminate_noop_with_shape ... 0.11ms 鈹屸攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹 鈹?鈹溾攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹尖攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹尖攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹 鈹?鈹?鈹?鈹溾攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹尖攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹尖攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹 鈹?鈹?鈹?鈹?鈹?鈹溾攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹尖攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹尖攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹 鈹?鈹?鈹?鈹?鈹?鈹?鈹?鈹?鈹?鈹?鈹?鈹斺攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹粹攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹粹攢鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹€鈹 ############# pass_level1 onnx ############# pass_level2 ############# pass_level3 open failed ############# pass_level4 ############# pass_level5 ############# pass_ncnn