Merge pull request #2 from Ahwar/docs-and-refactor

ref: Docs and Refactor Improvements

Author: Ahwar

.vscode/launch.json

@@ -13,7 +13,8 @@
  "python": "D:/ahar/Computer-Pointer-Controller-with-Eyes/.venv/Scripts/python.exe",
  "args": [
  "-i",
- "bin/face.png",
+ // "bin/face.png", // for single image
+ "CAM",
  "-ftm",
  "bin/models/face-detection-retail-0004.xml",
  "-ldm",

README.md

@@ -0,0 +1,12 @@
+curl --create-dirs https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/face-detection-retail-0004/FP32/face-detection-retail-0004.xml -o bin/models/face-detection-retail-0004.xml
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/face-detection-retail-0004/FP32/face-detection-retail-0004.bin -o bin/models/face-detection-retail-0004.bin
+
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/landmarks-regression-retail-0009/FP32/landmarks-regression-retail-0009.xml -o bin/models/landmarks_regression_retail_0009.xml
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/landmarks-regression-retail-0009/FP32/landmarks-regression-retail-0009.bin -o bin/models/landmarks_regression_retail_0009.bin
+
+
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/head-pose-estimation-adas-0001/FP32/head-pose-estimation-adas-0001.xml -o bin/models/head-pose-estimation-adas-0001.xml
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/head-pose-estimation-adas-0001/FP32/head-pose-estimation-adas-0001.bin -o bin/models/head-pose-estimation-adas-0001.bin
+
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/gaze-estimation-adas-0002/FP32/gaze-estimation-adas-0002.xml -o bin/models/gaze-estimation-adas-0002.xml
+curl https://storage.openvinotoolkit.org/repositories/open_model_zoo/2023.0/models_bin/1/gaze-estimation-adas-0002/FP32/gaze-estimation-adas-0002.bin -o bin/models/gaze-estimation-adas-0002.bin

download_models.sh

@@ -0,0 +1,72 @@
+import cv2
+import numpy as np
+import openvino as ov
+
+"""
+This is a sample class for a model. You may choose to use it as-is or make any changes to it.
+"""
+
+
+class CVModel:
+ """
+ Super Class for Running Computer vision Models.
+ """
+
+ def __init__(self, model_name, device="CPU"):
+ self.core = None
+ self.compiled_model = None
+ self.infer_request = None
+ self.device = device
+ self.model_xml = model_name
+
+ def load_model(self):
+ """
+ Load Model file and create Executable Network
+
+ This method is for loading the model to the device specified by the user.
+ If your model requires any Plugins (e.g. CPU extensions), this is where you can load them.
+ Initialize Core
+ Read model in to IENetwork and add any necessary Extension
+ Check for Supported Layers
+ """
+ ### Load the model ###
+ # Take model .xml and .bin file and create IENetwork
+ self.core = ov.Core()
+ self.compiled_model = self.core.compile_model(self.model_xml, "AUTO")
+
+ self.infer_request = self.compiled_model.create_infer_request()
+
+ def preprocess_input(self, image):
+ """
+ Before feeding the data into the model for inference,
+ you might have to preprocess it. This function is where you can do that.
+ """
+ (b, c, h, w) = self.get_input_shape()
+ image = cv2.resize(image, (w, h))
+ image = np.transpose(image, (2, 0, 1))
+ image = np.expand_dims(image, axis=0)
+ image = image.astype(np.float32)
+
+ return image
+
+ def predict(self, image):
+ """
+ Perform Inference on Image and return Raw Output.
+
+ This method is meant for running predictions on the input image.
+ Parameters:
+ image (numpy.ndarray): Frame from input file
+ """
+
+ ### PreProcess input image according to model Requirement
+ input_img = self.preprocess_input(image)
+ input_tensor = ov.Tensor(input_img, shared_memory=False)
+ ### run inference and return output
+ # Start Async Inference Request
+ self.infer_request.set_input_tensor(input_tensor)
+ self.infer_request.start_async()
+ self.infer_request.wait()
+
+ def get_input_shape(self):
+ """Return the shape of the input layer"""
+ return self.compiled_model.inputs[0].shape

lib/CVModel.py

@@ -1,80 +1,21 @@
 import cv2
-import numpy as np
 import openvino as ov
 
+from lib.CVModel import CVModel
+
 """
 This is a sample class for a model. You may choose to use it as-is or make any changes to it.
 """
 
 
-class FaceDetector:
- """
- Class for the Face Detection Model.
- """
-
- def __init__(self, model_name, device="CPU"):
- """Use this to set your instance variables."""
- self.core = None
- self.compiled_model = None
- self.infer_request = None
- self.device = device
- self.model_xml = model_name
-
- def load_model(self):
- """
- Load Model file and create Executable Network
-
- This method is for loading the model to the device specified by the user.
- If your model requires any Plugins (e.g. CPU extensions), this is where you can load them.
- Initialize Core
- Read model in to IENetwork and add any necessary Extension
- Check for Supported Layers
- """
- ### Load the model ###
- # Take model .xml and .bin file and create IENetwork
- self.core = ov.Core()
- self.compiled_model = self.core.compile_model(self.model_xml, "AUTO")
-
- self.infer_request = self.compiled_model.create_infer_request()
+class FaceDetector(CVModel):
 
  def predict(self, image):
- """
- Perform Inference on Image and return Raw Output.
-
- This method is meant for running predictions on the input image.
- Parameters:
- image (numpy.ndarray): Frame from input file
-
- Returns:
- model_output (numpy.ndarray): Raw Model Output
- """
-
- ### PreProcess input image according to model Requirement
- input_img = self.preprocess_input(image)
- input_tensor = ov.Tensor(input_img, shared_memory=False)
- ### run inference and return output
- # Start Async Inference Request
- self.infer_request.set_input_tensor(input_tensor)
- self.infer_request.start_async()
- self.infer_request.wait()
+ super().predict(image=image)
 
  model_output = self.infer_request.get_output_tensor().data
-
  return model_output
 
- def preprocess_input(self, image):
- """
- Before feeding the data into the model for inference,
- you might have to preprocess it. This function is where you can do that.
- """
- (b, c, h, w) = self.get_input_shape()
- image = cv2.resize(image, (w, h))
- image = np.transpose(image, (2, 0, 1))
- image = np.expand_dims(image, axis=0)
- image = image.astype(np.float32)
-
- return image
-
  def postprocess_output(self, outputs, threshold, image, image_w, image_h):
  """
  Before feeding the output of this model to the next model,
@@ -101,7 +42,7 @@ def get_input_shape(self):
 
 
 if __name__ == "__main__":
- detector = FaceDetector("bin/models/1/face-detection-retail-0004.xml")
+ detector = FaceDetector("bin/models/face-detection-retail-0004.xml")
  detector.load_model()
  image = cv2.imread("bin/face.png")
  print("Model Input shape:", detector.get_input_shape())

lib/face_detector.py

@@ -1,43 +1,15 @@
-import cv2
 import numpy as np
 import openvino as ov
 import math
 
-"""
-This is a sample class for a model. You may choose to use it as-is or make any changes to it.
-"""
+from lib.CVModel import CVModel
 
 
-class GazeEstimator:
+class GazeEstimator(CVModel):
  """
  Class for the Gaze Estimation Model.
  """
 
- def __init__(self, model_name, device="CPU"):
- """Use this to set your instance variables."""
- self.core = None
- self.compiled_model = None
- self.infer_request = None
- self.device = device
- self.model_xml = model_name
-
- def load_model(self):
- """
- Load Model file and create Executable Network
-
- This method is for loading the model to the device specified by the user.
- If your model requires any Plugins (e.g. CPU extensions), this is where you can load them.
- Initialize Core
- Read model in to IENetwork and add any necessary Extension
- Check for Supported Layers
- """
- ### Load the model ###
- # Take model .xml and .bin file and create IENetwork
- self.core = ov.Core()
- self.compiled_model = self.core.compile_model(self.model_xml, "AUTO")
-
- self.infer_request = self.compiled_model.create_infer_request()
-
  def predict(self, left_eye, right_eye, head_pose):
  """
  Perform Inference on Image and return Raw Output.
@@ -67,7 +39,6 @@ def predict(self, left_eye, right_eye, head_pose):
  dtype=np.float32,
  )
  poses = ov.Tensor(array=poses, shared_memory=False)
- print("++++", poses.shape)
 
  self.infer_request.set_input_tensors(
  {
@@ -87,19 +58,6 @@ def predict(self, left_eye, right_eye, head_pose):
  output_buffer = output.data
  return output_buffer
 
- def preprocess_input(self, image):
- """
- Before feeding the data into the model for inference,
- you might have to preprocess it. This function is where you can do that.
- """
- (b, c, h, w) = self.get_input_shape()
- image = cv2.resize(image, (w, h))
- image = np.transpose(image, (2, 0, 1))
- image = np.expand_dims(image, axis=0)
- image = image.astype(np.float32)
-
- return image
-
  def preprocess_output(self, outputs, hpe_cords):
  """
  Model output is dictionary like this
@@ -121,7 +79,3 @@ def preprocess_output(self, outputs, hpe_cords):
  except Exception as e:
  print("Error While preprocessing output in Gaze Estimation Model" + str(e))
  return mouse_cord
-
- def get_input_shape(self):
- """Return the shape of the input layer"""
- return self.compiled_model.inputs[0].shape

lib/gaze_estimator.py

@@ -1,42 +1,15 @@
-import cv2
-import numpy as np
-import openvino as ov
+from lib.CVModel import CVModel
 
 """
 This is a sample class for a model. You may choose to use it as-is or make any changes to it.
 """
 
 
-class HeadposeEstimator:
+class HeadposeEstimator(CVModel):
  """
  Class for the Face Detection Model.
  """
 
- def __init__(self, model_name, device="CPU"):
- """Use this to set your instance variables."""
- self.core = None
- self.compiled_model = None
- self.infer_request = None
- self.device = device
- self.model_xml = model_name
-
- def load_model(self):
- """
- Load Model file and create Executable Network
-
- This method is for loading the model to the device specified by the user.
- If your model requires any Plugins (e.g. CPU extensions), this is where you can load them.
- Initialize Core
- Read model in to IENetwork and add any necessary Extension
- Check for Supported Layers
- """
- ### Load the model ###
- # Take model .xml and .bin file and create IENetwork
- self.core = ov.Core()
- self.compiled_model = self.core.compile_model(self.model_xml, "AUTO")
-
- self.infer_request = self.compiled_model.create_infer_request()
-
  def predict(self, image):
  """
  Perform Inference on Image and return Raw Output.
@@ -48,34 +21,9 @@ def predict(self, image):
  Returns:
  model_output (numpy.ndarray): Raw Model Output
  """
+ super().predict(image)
  model_output = {}
- ### PreProcess input image according to model Requirement
- input_img = self.preprocess_input(image)
- input_tensor = ov.Tensor(input_img, shared_memory=False)
- ### run inference and return output
- # Start Async Inference Request
- self.infer_request.set_input_tensor(input_tensor)
- self.infer_request.start_async()
- self.infer_request.wait()
-
  model_output["yaw"] = self.infer_request.get_output_tensor(0).data[0][0]
  model_output["pitch"] = self.infer_request.get_output_tensor(1).data[0][0]
  model_output["role"] = self.infer_request.get_output_tensor(2).data[0][0]
  return model_output
-
- def preprocess_input(self, image):
- """
- Before feeding the data into the model for inference,
- you might have to preprocess it. This function is where you can do that.
- """
- (b, c, h, w) = self.get_input_shape()
- image = cv2.resize(image, (w, h))
- image = np.transpose(image, (2, 0, 1))
- image = np.expand_dims(image, axis=0)
- image = image.astype(np.float32)
-
- return image
-
- def get_input_shape(self):
- """Return the shape of the input layer"""
- return self.compiled_model.inputs[0].shape