Closed thongvhoang closed 3 years ago
I use this code for converting full int8 TFLite model to Edge TPU by following link: https://github.com/zldrobit/yolov5/tree/tf-android-tfl-detect: Code:
!edgetpu_compiler -s -a -o edgetpu $yolov5s_int8_path
Bug error generate by this code:
Edge TPU Compiler version 15.0.340273435 Model compiled successfully in 2375 ms. Input model: /content/drive/MyDrive/yolov5/runs/train/yolov5s_results/weights/best-int8.tflite Input size: 7.34MiB Output model: edgetpu/best-int8_edgetpu.tflite Output size: 7.84MiB On-chip memory used for caching model parameters: 6.73MiB On-chip memory remaining for caching model parameters: 6.75MiB Off-chip memory used for streaming uncached model parameters: 411.38KiB Number of Edge TPU subgraphs: 12 Total number of operations: 255 Operation log: edgetpu/best-int8_edgetpu.log Model successfully compiled but not all operations are supported by the Edge TPU. A percentage of the model will instead run on the CPU, which is slower. If possible, consider updating your model to use only operations supported by the Edge TPU. For details, visit g.co/coral/model-reqs. Number of operations that will run on Edge TPU: 235 Number of operations that will run on CPU: 20 Operator Count Status RESIZE_NEAREST_NEIGHBOR 2 Operation version not supported CONCATENATION 14 Mapped to Edge TPU RESHAPE 3 Mapped to Edge TPU TRANSPOSE 3 Operation not supported MUL 59 Mapped to Edge TPU STRIDED_SLICE 4 Only Strided-Slice with unitary strides supported LOGISTIC 51 Mapped to Edge TPU LEAKY_RELU 8 Operation not supported PAD 6 Mapped to Edge TPU ADD 15 Mapped to Edge TPU MAX_POOL_2D 3 Mapped to Edge TPU QUANTIZE 14 Mapped to Edge TPU QUANTIZE 3 Operation is otherwise supported, but not mapped due to some unspecified limitation CONV_2D 70 Mapped to Edge TPU Error opening file for writing: edgetpu/best-int8_edgetpu.tflite Internal compiler error. Aborting!
In additional, I run the code:
edgetpu_compiler -a $yolov5s_int8_path
-> Successfully. But deploy mobile when running app on Android Studio, appears the bug error:E/AndroidRuntime: FATAL EXCEPTION: inference Process: org.tensorflow.lite.examples.detection, PID: 21132 java.lang.RuntimeException: java.lang.IllegalStateException: Internal error: Unexpected failure when preparing tensor allocations: Encountered unresolved custom op: edgetpu-custom-op. Node number 0 (edgetpu-custom-op) failed to prepare.
My environment: I use Google Colab (Ubuntu 18.04), Tensorflow 2.4.0 Object Detection Model: YOLOv5s-int8
In conclusion, I want to ask how to fix that error "Error opening file for writing" Thank you.
I fixed the error "Error opening file for writing": Option 1: Remove "edgetpu" Option 2: Create folder edgetpu
@thongvhoang Can this issue be closed now ?
@thongvhoang Can this issue be closed now ?
Yes, it can.
I use this code for converting full int8 TFLite model to Edge TPU by following link: https://github.com/zldrobit/yolov5/tree/tf-android-tfl-detect: Code:
!edgetpu_compiler -s -a -o edgetpu $yolov5s_int8_path
Bug error generate by this code:
In additional, I run the code:
edgetpu_compiler -a $yolov5s_int8_path
-> Successfully. But deploy mobile when running app on Android Studio, appears the bug error:E/AndroidRuntime: FATAL EXCEPTION: inference Process: org.tensorflow.lite.examples.detection, PID: 21132 java.lang.RuntimeException: java.lang.IllegalStateException: Internal error: Unexpected failure when preparing tensor allocations: Encountered unresolved custom op: edgetpu-custom-op. Node number 0 (edgetpu-custom-op) failed to prepare.
My environment: I use Google Colab (Ubuntu 18.04), Tensorflow 2.4.0 Object Detection Model: YOLOv5s-int8
In conclusion, I want to ask how to fix that error above "Error opening file for writing" and "E/AndroidRuntime: FATAL EXCEPTION: inference" Thank you.