"ValueError: array of sample points is empty" when using object_detection/evaluation

[mokoe004]

When running evaluation operation mode with a YOLOv8 quantized model on a dataset in Darknet format (converted to .tsf), I encounter a crash during metrics calculation due to an empty array of sample points.

• The dataset contains only the "person" class.
• Model is from: stm-hotspot/ultralytics
• The model works in other operation modes like deployment and prediction.

Configuration file:

general:
   model_path: ./pretrained_models/yolov8n_416_quant_pc_uf_od_coco-person-st.tflite
   model_type: yolo_v8
operation_mode: prediction

prediction:
  test_files_path: ./datasets/voc2012_person/test_files
  target: host # host, stedgeai_host, stedgeai_n6

evaluation:
  profile: profile_O3
  input_type: uint8     # int8 / uint8 / float32
  output_type: int8     # int8 / uint8 / float32
  input_chpos: chlast   # chlast / chfirst
  output_chpos: chlast  # chlast / chfirst
  target: host # host, stedgeai_host, stedgeai_n6

dataset:
  class_names: [person]
  test_path: ./datasets/voc2012_person/mini


preprocessing:
  rescaling: { scale: 1/255, offset: 0 }
  resizing:
    aspect_ratio: fit
    interpolation: nearest
  color_mode: rgb

postprocessing:
  confidence_thresh: 0.4
  NMS_thresh: 0.5
  IoU_eval_thresh: 0.5
  plot_metrics: True   # Plot precision versus recall curves. Default is False.
  max_detection_boxes: 100

tools:
   stedgeai:
      path_to_stedgeai: C:/ST/STEdgeAI/2.1/Utilities/windows/stedgeai.exe

mlflow:
   uri: ./src/experiments_outputs/mlruns

hydra:
   run:
      dir: ./src/experiments_outputs/${now:%Y_%m_%d_%H_%M_%S}

Error log:

Error executing job with overrides: []
Traceback (most recent call last):
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\venv\lib\site-packages\clearml\binding\hydra_bind.py", line 230, in _patched_task_function
    return task_function(a_config, *a_args, **a_kwargs)
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\object_detection\stm32ai_main.py", line 220, in main
    process_mode(cfg)
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\object_detection\stm32ai_main.py", line 69, in process_mode
    evaluate(cfg)
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\object_detection\src\evaluation\evaluate.py", line 353, in evaluate
    metrics = _evaluate_quantized_model(cfg, model_path, num_classes=num_classes, output_dir=output_dir)
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\object_detection\src\evaluation\evaluate.py", line 199, in _evaluate_quantized_model
    metrics = calculate_objdet_metrics(groundtruths, detections, cpp.IoU_eval_thresh)
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\object_detection\src\utils\objdet_metrics.py", line 291, in calculate_objdet_metrics
    irec = np.interp(-x, -conf, rec, left=0)
  File "<__array_function__ internals>", line 180, in interp
  File "C:\Users\ibrak\PycharmProjects\stm32ai-modelzoo-services\venv\lib\site-packages\numpy\lib\function_base.py", line 1594, in interp
    return interp_func(x, xp, fp, left, right)
ValueError: array of sample points is empty