techshoww
yolov5-6.0转成的mnn是4个outputTensor怎么弄呢
我用yolov5-5.0转出的是3个outputTensor可以正常使用，但是换成yolov5-6.0就是4个输出outputTensor这个源码就不能用，请问怎么改成它能用呢。我用是yolov5 P5模型，按道理转出来mnn是3个outputTensor才对，但是实际出来是4个，而且也没报错。以下是6.0转换的日志： python -m MNN.tools.mnnconvert -f ONNX --modelFile C:\Users\fut\Desktop\insulator_5cls_yolov5s_6.0\best.onnx --MNNModel C:\Users\fut\Desktop\insulator_5cls_yolov5s_6.0\best.mnn The device support dot:0, support fp16:0, support i8mm: 0 The device support dot:0, support fp16:0, support i8mm: 0 D:\anaconda3\lib\runpy.py:127: RuntimeWarning: 'MNN.tools.mnnconvert' found in sys.modules after import of package 'MNN.tools', but prior to execution of 'MNN.tools.mnnconvert'; this may result in unpredictable behaviour warn(RuntimeWarning(msg)) Don't has bizCode, use MNNTest for default Start to Convert Other Model Format To MNN Model... [13:21:05] @ 40: ONNX Model ir version: 6 Start to Optimize the MNN Net... inputTensors : [ images, ] outputTensors: [ 345, 403, 461, output, ] Converted Success! 可以看到是4个输出
