Multi-TPU fails with MobileNet large

[Daniel15]

CodeProject.AI Server info

Server version:   2.9.7
System:           Docker (b8c4699aeb25)
Operating System: Linux (Ubuntu 22.04 Jammy Jellyfish)
CPUs:             13th Gen Intel(R) Core(TM) i5-13500 (Intel)
                  1 CPU x 14 cores. 20 logical processors (x64)
System RAM:       125 GiB
Platform:         Linux
BuildConfig:      Release
Execution Env:    Docker
Runtime Env:      Production
Runtimes installed:
  .NET runtime:     9.0.0
  .NET SDK:         Not found
  Default Python:   3.10.12
  Go:               Not found
  NodeJS:           Not found
  Rust:             Not found
Video adapter info:
System GPU info:
  GPU 3D Usage       0%
  GPU RAM Usage      0
Global Environment variables:
  CPAI_APPROOTPATH = <root>
  CPAI_PORT        = 32168

Describe the bug
When using the MobileNet large model (and possibly others) and the "Multi-TPU Support" option is enabled, it fails to initialize:

17:07:00:objectdetection_coral_adapter.py: ERROR:root:TFLite file /app/modules/ObjectDetectionCoral/assets/all_segments_tf2_ssd_mobilenet_v1_fpn_640x640_coco17_ptq_segment_0_of_2_edgetpu.tflite doesn't exist
17:07:00:objectdetection_coral_adapter.py: WARNING:root:Model file not found: [Errno 2] No such file or directory: '/app/modules/ObjectDetectionCoral/assets/all_segments_tf2_ssd_mobilenet_v1_fpn_640x640_coco17_ptq_segment_0_of_2_edgetpu.tflite'
17:07:00:objectdetection_coral_adapter.py: WARNING:root:No Coral TPUs found or able to be initialized. Using CPU.
17:07:00:objectdetection_coral_adapter.py: WARNING:root:Unable to load delegate for TPU cpu: Failed to load delegate from libedgetpu.so.1
17:07:00:objectdetection_coral_adapter.py: WARNING:root:Unable to create interpreter for CPU using edgeTPU library: cpu
17:07:00:objectdetection_coral_adapter.py: Using PIL for image manipulation (Either OpenCV or numpy not available for this module)
17:07:00:objectdetection_coral_adapter.py: TPU detected
17:07:00:objectdetection_coral_adapter.py: Attempting multi-TPU initialisation
17:07:00:objectdetection_coral_adapter.py: Failed to init multi-TPU. Falling back to single TPU.
17:07:00:objectdetection_coral_adapter.py: Using Edge TPU

Falling back to a single TPU works fine, and the single TPU also initializes correctly when Multi-TPU support is disabled.

To Reproduce
Steps to reproduce the behavior:

1. Install dual-edge TPU
2. Enable multi-TPU mode
3. Download MobileNet Large model
4. Use MobileNet Large model

Expected behavior
It works with no errors

[MatrixTechno]

Im having the exact same issue with YOLOv5 and v8 no matter the size