Quickstart for beginners

This short introduction use kp to build simple edge AI application by following steps:

1. Connect Kneron device
2. Upload SCPU/NCPU firmware
3. Upload NEF model
4. Inference image

Note 1: Please using pip to install kp API and opencv-python before the following tutorial. See the Installation for details.

Note 2: Please using kneronDFUT to upgrade firmware from KDP to KDP2 before the following tutorial. See the Upgrade AI Device to KDP2 for details.

---

Import kp and cv2 into your program:

import kp
import cv2

---

Get one Kneron device USB port ID for connecting:

device_descriptors = kp.core.scan_devices()

if 0 < device_descriptors.device_descriptor_number:
    usb_port_id = device_descriptors.device_descriptor_list[0].usb_port_id
else:
    print('Error: no Kneron device connect.')
    exit(0)

---

Connect the device and get Kneron device handler (Device Group):

device_group = kp.core.connect_devices(usb_port_ids=[usb_port_id])

---

Set timeout of the USB communication (Default: infinity wait):

kp.core.set_timeout(device_group=device_group,
                    milliseconds=5000)

---

Upload firmware to Kneron device:

Please replace SCPU_FW_PATH, NCPU_FW_PATH by fw_scpu.bin and fw_ncpu.bin path (Please find target device firmware under res/firmware folder)

SCPU_FW_PATH = 'res/firmware/KL520/fw_scpu.bin'
NCPU_FW_PATH = 'res/firmware/KL520/fw_ncpu.bin'
kp.core.load_firmware_from_file(device_group=device_group,
                                scpu_fw_path=SCPU_FW_PATH,
                                ncpu_fw_path=NCPU_FW_PATH)

---

Upload NEF model to Kneron device:

Please replace MODEL_FILE_PATH by models_520.nef path (Please find target device NEF model under res/models folder)

MODEL_FILE_PATH = 'res/models/KL520/tiny_yolo_v3/models_520.nef'
model_nef_descriptor = kp.core.load_model_from_file(device_group=device_group,
                                                    file_path=MODEL_FILE_PATH)

---

Inference image:

• Read image from disk

  Please replace IMAGE_FILE_PATH by image path (Example image can be found under res/images folder)

  IMAGE_FILE_PATH = 'res/images/bike_cars_street_224x224.bmp'
  
  img = cv2.imread(filename=IMAGE_FILE_PATH)
  img_bgr565 = cv2.cvtColor(src=img, code=cv2.COLOR_BGR2BGR565)

• Prepare generic image inference input descriptor

  generic_inference_input_descriptor = kp.GenericImageInferenceDescriptor(
      model_id=model_nef_descriptor.models[0].id,
      inference_number=0,
      input_node_image_list=[
          kp.GenericInputNodeImage(
              image=img_bgr565,
              image_format=kp.ImageFormat.KP_IMAGE_FORMAT_RGB565,
              resize_mode=kp.ResizeMode.KP_RESIZE_ENABLE,
              padding_mode=kp.PaddingMode.KP_PADDING_CORNER,
              normalize_mode=kp.NormalizeMode.KP_NORMALIZE_KNERON
          )
      ]
  )

• Start inference work

  kp.inference.generic_image_inference_send(device_group=device_group,
                                            generic_inference_input_descriptor=generic_inference_input_descriptor)
  generic_raw_result = kp.inference.generic_image_inference_receive(device_group=device_group)

• Retrieve inference node output

  inf_node_output_list = []
  
  for node_idx in range(generic_raw_result.header.num_output_node):
      inference_float_node_output = kp.inference.generic_inference_retrieve_float_node(node_idx=node_idx,
                                                                                       generic_raw_result=generic_raw_result,
                                                                                       channels_ordering=kp.ChannelOrdering.KP_CHANNEL_ORDERING_CHW)
      inf_node_output_list.append(inference_float_node_output)
  
  print(inf_node_output_list)
  
  '''
  [{
      "width": 7,
      "height": 7,
      "channel": 255,
      "channels_ordering": "ChannelOrdering.KP_CHANNEL_ORDERING_CHW",
      "num_data": 12495,
      "ndarray": [
          "[[[[  1.3589103    0.33972758   0.5095914  ...   0.16986379",
          "      0.33972758  -0.849319  ]",
          "   [  1.698638    -0.5095914    0.5095914  ...  -0.16986379",
          "     -0.16986379  -0.849319  ]",
          "   [  1.5287741    0.5095914    0.16986379 ...   0.",
          "     -0.33972758  -0.5095914 ]",
          "   ...",
          "   [ -7.643871    -9.682237   -10.361691   ...  -9.002781",
          "    -10.021964    -9.172645  ]",
          "   [ -7.304143   -10.701419   -12.400057   ...  -9.682237",
          "     -9.682237    -9.002781  ]",
          "   [ -6.1150966   -8.153462    -9.342508   ...  -7.983598",
          "     -8.153462    -8.49319   ]]]]"
      ]
  }, {
      "width": 14,
      "height": 14,
      "channel": 255,
      "channels_ordering": "ChannelOrdering.KP_CHANNEL_ORDERING_CHW",
      "num_data": 49980,
      "ndarray": [
          "[[[[  0.87369454  -0.3494778   -0.1747389  ...   0.",
          "     -0.1747389   -0.6989556 ]",
          "   [  0.6989556   -0.87369454  -0.6989556  ...  -0.5242167",
          "     -0.1747389   -0.5242167 ]",
          "   [  0.5242167   -0.87369454  -0.6989556  ...  -0.1747389",
          "      0.1747389   -0.87369454]",
          "   ...",
          "   [-11.18329    -15.377024   -18.172846   ... -13.105417",
          "    -12.581201   -10.833812  ]",
          "   [-10.134856   -14.1538515  -16.774935   ... -10.659073",
          "     -9.61064     -8.387467  ]",
          "   [ -9.086423   -12.231723   -12.930678   ... -10.134856",
          "     -9.261162    -7.339034  ]]]]"
      ]
  }]
  '''