feat: Implement basic face detection inference

Demonstrates OpenVINO inference on a sample image using the face-detection-retail-0004 model.

Author: Ahwar

.gitignore

@@ -1,3 +1,5 @@
 src/__pycache__
-models
-main.log
+main.log
+bin/models
+
+venv

requirements.txt

@@ -1,11 +1,3 @@
-image==1.5.27
-ipdb==0.12.3
-ipython==7.10.2
-numpy==1.17.4
-Pillow==6.2.1
-requests==2.22.0
-virtualenv==16.7.9
-pyyaml==5.3.1
-opencv-python==4.3.0.36
-PyAutoGUI==0.9.50
-python3-xlib==0.15
+openvino
+opencv-python
+black

sample inference.py

@@ -0,0 +1,53 @@
+"""
+Just a sample code that demonstrate inference using OpenVino
+"""
+
+import openvino as ov
+import cv2
+import numpy as np
+
+## download models
+# curl --create-dirs https://storage.openvinotoolkit.org/repositories/open_model_zoo/2022.1/models_bin/2/face-detection-retail-0004/FP32/face-detection-retail-0004.bin -o model/1/face-detection-retail-0004.bin
+# curl --create-dirs https://storage.openvinotoolkit.org/repositories/open_model_zoo/2022.1/models_bin/2/face-detection-retail-0004/FP32/face-detection-retail-0004.xml -o model/1/face-detection-retail-0004.xml
+
+core = ov.Core()
+
+compiled_model = core.compile_model(r"model\1\face-detection-retail-0004.xml", "AUTO")
+
+infer_request = compiled_model.create_infer_request()
+
+print(infer_request.get_input_tensor())
+print(infer_request.get_output_tensor())
+
+
+## read image and preprocessing
+# Read image
+img = cv2.imread(r"bin\face.png")
+# Resize image
+img = cv2.resize(img, (300, 300))
+# Convert to float32
+img = img.astype(np.float32)
+# Normalize pixel values to [0, 1]
+img = img / 255.0
+# Convert to numpy array
+img = np.array(img)
+img = img.reshape(3, 300, 300)
+img = np.expand_dims(img, axis=0)
+
+
+## Inference
+# Create tensor from external memory
+input_tensor = ov.Tensor(array=img, shared_memory=True)
+# Set input tensor for model with one input
+infer_request.set_input_tensor(input_tensor)
+# run inference
+infer_request.start_async()
+infer_request.wait()
+
+
+## output post processing
+# get output
+# Get output tensor for model with one output
+output = infer_request.get_output_tensor()
+output_buffer = output.data
+print(output_buffer.shape)