problems with detection models

[ismukhin]

Проблема заключается в следующем:

1. модель detr-resnet50 выдает ошибку на тестовом изображении (где изображены компьютер, мышь, клавиатура):

(openvino-venv) PS C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\src\inference> python .\inference_sync_mode.py -m ..\..\open_model_zoo\public\detr-resnet50\FP32\detr-resnet50.xml -w ..\..\open_model_zoo\public\detr-resnet50\FP32\detr-resnet50.bin -i ..\..\coco1.jpg -t detection
[ INFO ] Inference Engine initialization
[ INFO ] Loading model files:
         ..\..\open_model_zoo\public\detr-resnet50\FP32\detr-resnet50.xml
         ..\..\open_model_zoo\public\detr-resnet50\FP32\detr-resnet50.bin
[ INFO ] Shape for input layer input: 1x3x800x1137
[ INFO ] Prepare input data
[ INFO ] Create executable network
[ INFO ] Starting inference (1 iterations) on CPU
ERROR! : invalid index to scalar variable.

2. модель yolo-v4-tf выдает отрицательные значения окаймляющих прямоугольников и в конце выдает ошибку:

[ INFO ] Inference Engine initialization
[ INFO ] Loading model files:
         ..\..\open_model_zoo\public\yolo-v4-tf\FP32\yolo-v4-tf.xml
         ..\..\open_model_zoo\public\yolo-v4-tf\FP32\yolo-v4-tf.bin
[ INFO ] Shape for input layer Func/StatefulPartitionedCall/input/_0:0: 1x608x608x3
[ INFO ] Prepare input data
[ INFO ] Create executable network
[ INFO ] Starting inference (1 iterations) on CPU
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-680, -5551)
[ INFO ] Bottom right: (-2053, -4092)
[ INFO ] Bounding boxes for image 0 for object 1
[ INFO ] Top left: (-642, -6705)
[ INFO ] Bottom right: (-2252, -4626)
[ INFO ] Bounding boxes for image 0 for object 1
[ INFO ] Top left: (-620, -8023)
[ INFO ] Bottom right: (-2518, -4884)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-597, -8462)
[ INFO ] Bottom right: (-3137, -5240)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-585, -8322)
[ INFO ] Bottom right: (-3384, -5328)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-644, -8560)
[ INFO ] Bottom right: (-2738, -5025)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-732, -8355)
[ INFO ] Bottom right: (-2621, -4641)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-781, -8350)
[ INFO ] Bottom right: (-2005, -4456)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-769, -8009)
[ INFO ] Bottom right: (-1647, -4277)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-782, -8150)
[ INFO ] Bottom right: (-1312, -4214)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-765, -8033)
[ INFO ] Bottom right: (-1234, -4298)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-774, -7875)
[ INFO ] Bottom right: (-1220, -4222)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-772, -7701)
[ INFO ] Bottom right: (-1295, -4161)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-793, -7837)
[ INFO ] Bottom right: (-1368, -4031)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-767, -7980)
[ INFO ] Bottom right: (-1224, -3877)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-733, -7708)
[ INFO ] Bottom right: (-1528, -3924)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-730, -7396)
[ INFO ] Bottom right: (-2027, -4367)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-745, -7180)
[ INFO ] Bottom right: (-2622, -5203)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-739, -6894)
[ INFO ] Bottom right: (-3067, -5733)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-735, -6525)
[ INFO ] Bottom right: (-3351, -5910)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-715, -6819)
[ INFO ] Bottom right: (-3197, -5806)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-697, -7709)
[ INFO ] Bottom right: (-2912, -5807)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-659, -7713)
[ INFO ] Bottom right: (-3103, -5785)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-666, -7339)
[ INFO ] Bottom right: (-2931, -5441)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-734, -7343)
[ INFO ] Bottom right: (-2841, -5287)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-727, -6707)
[ INFO ] Bottom right: (-3433, -5921)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-683, -5781)
[ INFO ] Bottom right: (-4617, -6702)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-668, -6954)
[ INFO ] Bottom right: (-3207, -5577)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-684, -8097)
[ INFO ] Bottom right: (-1572, -4067)
[ INFO ] Bounding boxes for image 0 for object 0
[ INFO ] Top left: (-678, -7791)
[ INFO ] Bottom right: (-1320, -3818)
ERROR! : list index out of range

3. модели efficientdet-d0/1-tf не конвертируются:

(openvino-venv) PS C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\src\inference> omz_converter --name efficientdet-d0-tf --download_dir ..\..\open_model_zoo\
========== Running pre-convert script for efficientdet-d0-tf
Pre-convert command: C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\openvino-venv\Scripts\python.exe -- C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\openvino-venv\Lib\site-packages\openvino\model_zoo\models\public\efficientdet-d0-tf\pre-convert.py -- ..\..\open_model_zoo\public\efficientdet-d0-tf ..\..\open_model_zoo\public\efficientdet-d0-tf

2022-11-07 14:03:14.596251: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found
2022-11-07 14:03:14.596917: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your 
machine.
Traceback (most recent call last):
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\open_model_zoo\public\efficientdet-d0-tf\model\model_inspect.py", line 29, in <module>
    import inference
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\open_model_zoo\public\efficientdet-d0-tf\model\inference.py", line 26, in <module>
    import dataloader
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\open_model_zoo\public\efficientdet-d0-tf\model\dataloader.py", line 19, in <module>
    import utils
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\open_model_zoo\public\efficientdet-d0-tf\model\utils.py", line 204, in <module>
    class TpuBatchNormalization(tf.keras.layers.BatchNormalization):
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\openvino-venv\lib\site-packages\tensorflow\python\util\lazy_loader.py", line 58, in __getattr__
    module = self._load()
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\openvino-venv\lib\site-packages\tensorflow\python\util\lazy_loader.py", line 41, in _load
    module = importlib.import_module(self.__name__)
  File "C:\Users\Ivashka\.pyenv\pyenv-win\versions\3.9.12\lib\importlib\__init__.py", line 127, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
ModuleNotFoundError: No module named 'keras.api'
Traceback (most recent call last):
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\openvino-venv\Lib\site-packages\openvino\model_zoo\models\public\efficientdet-d0-tf\pre-convert.py", line 43, in <module>
    main()
  File "C:\Users\Ivashka\Desktop\Studying\BIG_DATA_and_ML\DL_Benchmark\dl-benchmark\openvino-venv\Lib\site-packages\openvino\model_zoo\models\public\efficientdet-d0-tf\pre-convert.py", line 34, in main
    subprocess.run([sys.executable, '--',
  File "C:\Users\Ivashka\.pyenv\pyenv-win\versions\3.9.12\lib\subprocess.py", line 528, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['C:\\Users\\Ivashka\\Desktop\\Studying\\BIG_DATA_and_ML\\DL_Benchmark\\dl-benchmark\\openvino-venv\\Scripts\\python.exe', '--', '..\\..\\open_model_zoo\\public\\efficientdet-d0-tf\\model\\model_inspect.py', '--runmode=saved_model', '--model_name=efficientdet-d0', '--ckpt_path=..\\..\\open_model_zoo\\public\\efficientdet-d0-tf\\efficientdet-d0', '--saved_model_dir=..\\..\\open_model_zoo\\public\\efficientdet-d0-tf\\efficientdet-d0_saved_model']' returned non-zero exit status 1.

FAILED:
efficientdet-d0-tf

Версии библиотек в файле lib.txt
lib.txt