Skip to content

An error while starting ObjectDetectionCoral with multi-file models #6

@serg987

Description

@serg987

Area of Concern

  • Server
  • Behaviour of ObjectDetectionCoral
  • Installer
  • Runtime [e.g. Python3.7, .NET]
  • Module packages [e.g. PyTorch)
  • Something else

Describe the bug
Steps to repro:
Download YOLOv5 Medium or YOLOv8 Medium Models through UI
Restart ObjectDetectionCoral

Actual behavior
Failed to load the model, cannot find asset files
Logs for YOLOv8 Medium

22:39:54:CPAI_CORAL_MODEL_NAME = YOLOv8
22:39:54:CPAI_CORAL_MULTI_TPU  = True
22:39:54:MODELS_DIR            = <root>/modules/ObjectDetectionCoral/assets
22:39:54:MODEL_SIZE            = medium
22:39:54:
22:39:54:Started Object Detection (Coral) module
22:39:55:objectdetection_coral_adapter.py: ERROR:root:TFLite file /app/modules/ObjectDetectionCoral/assets/yolov8m__segment_0_of_2_edgetpu.tflite doesn't exist
22:39:55:objectdetection_coral_adapter.py: WARNING:root:Model file not found: [Errno 2] No such file or directory: '/app/modules/ObjectDetectionCoral/assets/yolov8m__segment_0_of_2_edgetpu.tflite'
22:39:55:objectdetection_coral_adapter.py: WARNING:root:No Coral TPUs found or able to be initialized. Using CPU.
22:39:55:objectdetection_coral_adapter.py: ERROR:root:TFLite file /app/modules/ObjectDetectionCoral/assets/yolov8m-416_640px.tflite doesn't exist
22:39:55:objectdetection_coral_adapter.py: WARNING:root:Unable to create interpreter for CPU using edgeTPU library: [Errno 2] No such file or directory: '/app/modules/ObjectDetectionCoral/assets/yolov8m-416_640px.tflite'

Logs for YOLOv5 Medium:

23:04:00:CPAI_CORAL_MODEL_NAME = YOLOv5
23:04:00:CPAI_CORAL_MULTI_TPU  = True
23:04:00:MODELS_DIR            = <root>/modules/ObjectDetectionCoral/assets
23:04:00:MODEL_SIZE            = medium
23:04:00:
23:04:00:Started Object Detection (Coral) module
23:04:01:objectdetection_coral_adapter.py: ERROR:root:TFLite file /app/modules/ObjectDetectionCoral/assets/yolov5m-int8_edgetpu_segment_0_of_4_edgetpu.tflite doesn't exist
23:04:01:objectdetection_coral_adapter.py: WARNING:root:Model file not found: [Errno 2] No such file or directory: '/app/modules/ObjectDetectionCoral/assets/yolov5m-int8_edgetpu_segment_0_of_4_edgetpu.tflite'
23:04:01:objectdetection_coral_adapter.py: WARNING:root:No Coral TPUs found or able to be initialized. Using CPU.
23:04:01:objectdetection_coral_adapter.py: WARNING:root:Unable to load delegate for TPU cpu: Failed to load delegate from libedgetpu.so.1
23:04:01:objectdetection_coral_adapter.py: WARNING:root:Unable to create interpreter for CPU using edgeTPU library: cpu

Looks like there's a discrepancy between what the module wants and what is actually in assets;

there's no yolov8m__segment_0_of_2_edgetpu.tflite, but yolov8m_416_640px_segment_0_of_2_edgetpu.tflite

no yolov8m-416_640px.tflite, but yolov8m_416_640px.tflite

no yolov5m-int8_edgetpu_segment_0_of_4_edgetpu.tflite, but yolov5m-int8_segment_0_of_4.tflite

After manually renaming the files it works well.

Though interesting - as it wants yolov8m__segment_0_of_2_edgetpu.tflite for the first segment, for the second one it wants the existing yolov8m_416_640px_segment_1_of_2_edgetpu.tflite. For YOLOv5 all files were renamed to make it work.

Expected behavior
ObjectDetectionCoral should start with the multi-file model

Screenshots
YOLOv5 logs:
240418-2

YOLOv8 logs;
240418-1

ls assets/:
240418-3

Your System (please complete the following information):

  • CodeProject.AI Server version: 2.6.2 in Docker
  • OS: Debian 12
  • System RAM: 32Gb
  • no GPU
  • 4xCoral TPU PCIe

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions