AI Master Class series –

Day 11
Object Recognition
Day-11 Agenda.

01. 02. 03.

Implementing Object
Object Recognition Recognition Deep Neural Network
Object Recognition Methodology DNN & Object
recognition Model

04. 05.
Deploying Real-time Object
MobileNetSSD recognition
Pre-trained Model Pre-Trained Model
Object Recognition.
Object recognition is a computer vision technique
for identifying objects in images or videos. Object
recognition is a key output of deep learning and
machine learning algorithms. When humans look at
a photograph or watch a video, we can readily spot
people, objects, scenes, and visual details.
Implementing Object
Recognition.
PRE-TRAINED
0
1 MODEL

0
TRANSFER
2 LEARNING

0
BUILDING FROM
3 SCRATCH
Deep Neural Network -
DNN.
• Solve Complex Task
• When it gets new information in the system, it
learns how to act accordingly to a new situation.
• Learning becomes deeper when tasks you solve
get harder.
• Helps to load pre-trained Model from DL
frameworks such as
 Tensorflow
 Caffe
 Darknet
 Torch
Speed Comparison on Image
Classification.
Pre-trained Model for Object
recognition.
• MobileNet-SSD
• GoogleNet
• Squeezenet
• Faster R-CNN
• ResNet
• Inception
• YOLO
• VGGNet
MobileNet SSD (Single shot Multibox
Detector).
• The MobileNet model is based on depthwise separable convolutions which are a
form of factorized convolutions. These factorize a standard convolution into a
depthwise convolution and a 1 × 1 convolution called a pointwise convolution.

• For MobileNets, the depthwise convolution applies a single filter to each input
channel. The pointwise convolution then applies a 1 × 1 convolution to combine
the outputs of the depthwise convolution.

• A standard convolution both filters and combines inputs into a new set of outputs
in one step. The depthwise separable convolution splits this into two layers
– a separate layer for filtering and a separate layer for combining. This
factorization has the effect of drastically reducing computation and model size.

• The SSD architecture is a single convolution network that learns to predict

bounding box locations and classify these locations in one pass. Hence, SSD can be
trained end-to-end.
MobileNet SSD.
MobileNet SSD
Architecture.
ReLu.
• The Rectified Linear Unit is the most commonly
used activation function in deep learning
models.

• The function returns 0 if it receives any negative

input, but for any positive value x it returns
that value back. So it can be written as
f(x)=max(0,x) .

• the ReLu function is able to accelerate the

training speed of deep neural networks
compared to traditional activation functions
since the derivative of ReLu is 1 for a positive
input.

• Due to a constant, deep neural networks do not

need to take additional time for computing error
terms during training phase.
OpenCV Basic Syntax for DNN.
Loading Image from Disk to DNN
cv2.dnn.blobFromImage
cv2.dnn.blobFromImages

Import Model from various Framework

cv2.dnn.createCaffeImporter
cv2.dnn.createTensorFlowImporter
cv2.dnn.createTorchImporter

cv2.dnn.readNetFromCaffe
cv2.dnn.readNetFromTensorFlow
cv2.dnn.readNetFromTorch
cv2.dnn.readhTorchBlob

.forward` method is used to forward-propagate our image and obtain

cv2.dnn.blobFromImage.

Mean Subtracted Normalized Image =

cv2.dnn.blobFromImage(resizedImage,scalingFactor,
Spatial Size, Mean Subtraction Values)

blob =
cv2.dnn.blobFromImage(imResizeBlob,0.007843,
(300, 300), 127.5)
Block Diagram – Workflow of DNN in
OpenCV.
Select Reading frame
Load Model Select target
Backend from camera

Convert to
Forward Post Process
Blob
Numpy Basic Syntax.
Numpy.array
arr = np.array([1, 2, 3, 4, 5])

Numpy.arrange

NumPy arange() is one of the array creation routines

based on numerical ranges. It creates an instance of
ndarray with evenly spaced values and returns the
reference to it.

for i in np.arange(0,detShape):
 Practical
session using
Object Recognition
Pre-trained Model with DNN
in OpenCV
import numpy as np
import imutils
import time
import cv2

prototxt = "MobileNetSSD_deploy.prototxt.txt"
model = "MobileNetSSD_deploy.caffemodel"
confThresh = 0.2
CLASSES = ["background", "aeroplane", "bicycle", "bird", "boat",
"bottle", "bus", "car", "cat", "chair", "cow", "diningtable",
"dog", "horse", "motorbike", "person", "pottedplant", "sheep",
"sofa", "train", "tvmonitor"]
COLORS = np.random.uniform(0, 255, size=(len(CLASSES), 3))
print("Loading model...")
net = cv2.dnn.readNetFromCaffe(prototxt, model)
print("Model Loaded")
print("Starting Camera Feed...")
vs = cv2.VideoCapture(0)
time.sleep(2.0)
while True:
_,frame = vs.read()
frame = imutils.resize(frame, width=500)

(h, w) = frame.shape[:2]
imResizeBlob = cv2.resize(frame, (300, 300))
blob = cv2.dnn.blobFromImage(imResizeBlob,0.007843, (300, 300), 127.5)
net.setInput(blob)
detections = net.forward()
detShape = detections.shape[2]

for i in np.arange(0,detShape):
confidence = detections[0, 0, i, 2]
if confidence > confThresh:
idx = int(detections[0, 0, i, 1])
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
(startX, startY, endX, endY) = box.astype("int")
label = "{}: {:.2f}%".format(CLASSES[idx],confidence * 100)
cv2.rectangle(frame, (startX, startY), (endX, endY),COLORS[idx], 2)
if startY - 15 > 15:
y = startY - 15
else:
startY + 15
cv2.putText(frame, label, (startX, y),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, COLORS[idx], 2)
cv2.imshow("Frame", frame)
key = cv2.waitKey(1)
if key == 27:
break
vs.release()
cv2.destroyAllWindows()
AI News – Day 11.
Yesterday - 2020

American Institute of Physics

● The range of AI technologies available for
dealing with brain disease is growing fast,
and exciting new methods are being
applied to brain problems as computer
scientists gain a deeper understanding of
the capabilities of advanced algorithms.
● Researchers conducted a systematic
literature review to understand the state
of the art in the use of AI for brain
disease. Their qualitative review sheds
light on the most interesting corners of AI
development.
 Tomorrow
Thanks! session
Do you have any questions? Image Classification using
[email protected] CNN

www.pantechsolutions.net