E3S Web of Conferences 391, 01016 (2023) https://doi.org/10.
1051/e3sconf/202339101016
ICMED-ICMPC 2023
Object and it’s dimension detection in real time
Madhavi Karanam1, Varun Kumar Kamani1,*, Vikas Kuchana1, Gopal Krishna Reddy
Koppula1 , and Gautham Gongada1.
1Department of CSE, Gokaraju Rangaraju Institute of Engineering and Technology, India
Abstract. Object and its dimension detection from images and videos can
be very helpful for everyday use. This paper discusses the use of the
system to detect an object in real time and provide its dimensions upon
demand. The object dimension measurement and detection are some of the
important topics of computer vision which helps in automating the manual
tasks. Human beings are capable of recognizing and spotting objects in
images and videos, but computers lack that ability with out prior training.
To train the computer, we must use machine learning, computer vision, and
object detection algorithms. This project provides the way to detect and
measure an object’s dimension in real time from a webcam. To estimate
the object's dimension in real time, we have utilized the OpenCV and
NumPy libraries. Computer vision provides support to computers to
observe and understand. Computer vision helps the computer in
understanding a 3D surrounding from a 2D image and trains the computer
to perform different functions. It also helps in Human Computer
Interaction effectively because it is able to differentiate the objects with
surroundings and provide us with the key information.
1 Introduction
The study of how computers perceive and understand digital images and videos is called
computer vision. Computer vision embodies all of the activities provided by the biological
visual system, which includes "seeing" or interpreting visual stimuli, understanding what is
being observed, and distilling complex concepts into a format that will be used in a variety
of relevant and useful processes. This multidisciplinary field uses sensors, computers, and
machine learning techniques to imitate and automate these components of human visual
systems. Object Detection comes under Computer vision technology. Computer vision
provides the machine to have the ability to observe, understand, locate certain things in the
images or videos. Object Detection will also provide help to track the objects and we can
categorize the objects we have discovered and find examples of those items in the real time
scenario. There are several applications for object identification and tracking in computer
vision research, including traffic detection, vehicle navigation, and interpersonal
connections. object detection in digital photos and videos refers to the process of finding
examples of semantic items of a specific class utilizing computer equipment connected to
computer vision and image processing
© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons
Attribution License 4.0 (https://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 391, 01016 (2023) https://doi.org/10.1051/e3sconf/202339101016
ICMED-ICMPC 2023
Three of the numerous applications for object detection include face detection, face
recognition, and video objects. The object detection system detects whether things are
present or absent in particular scenes and from particular camera angles. The diverse object
detection domains are based on varied goals and are categorized using concrete and abstract
terms. The various models that are either expressly or implicitly employed for object
detection. The components may or may not vary depending on the different techniques,
selecting an item based on a hypothesis and selecting an item based on matching. The most
effective processing technique is object detection. Identifying objects in a situation is the
process of object detection in practical settings.
This paper aim is to propose an approach that can detect an object and provide its
dimensions. The measurements can be utilized in various applications, such as automatic
parking solutions, self-driving automobiles, and e-commerce platforms (such as Myntra,
Ajio, Amazon, Flipkart, etc.) to determine the size of the products to best suit the needs of
customers. The discussion of related work has been done in Section 2. Section 3 describes
the proposed approach, Section 4 presents the proposed approach's results, and Section 5
concludes the paper.
2 Related work
In this literature survey, we review various papers that explore different approaches for
object detection. An approach for real-time object detection and tracking using deep
learning with the help of OpenCV used a pre-trained deep neural network, namely You
Only Look Once (YOLO), which is designed to detect objects in real-time. The proposed
method achieved an accuracy of 92.6% on the PASCAL VOC 2012 data-set [1]. A review
of research on object detection based on deep learning surveyed recent advancements in
object detection, including Faster R-CNN, YOLO, SSD, and RetinaNet, it also discusses
various datasets used for object detection and the evaluation metrics used to measure the
performance of object detection algorithms [2]. The paper studied object detection and
proposed an approach that used a combination of image processing techniques and machine
learning algorithms to detect and classify objects. Various methods are researched in this,
and calculated accuracy of the methods. The final proposed method achieved an accuracy
of 87.5% on a custom datasets [3]. A comparative study of object detection algorithms,
including Haar cascades, Histogram of Oriented Gradients (HOG), and deep learning-based
methods such as YOLO, Faster R-CNN, and SSD. The study compared the accuracy,
speed, and computational complexity of these algorithms [4]. Another paper proposed an
efficient object detection algorithm that achieved state-of-the-art results on the COCO data-
set. The authors used a compound scaling method to balance accuracy and efficiency and
proposed a new scaling rule to determine the depth and width of the network [5].
A study of object detection methods and their applications on digital images. The
authors discussed various object detection algorithms, including traditional computer vision
methods such as Haar cascades and HOG, and deep learning-based methods such as YOLO
and Faster R-CNN [6]. Another study proposes a system for detecting and tracking objects
using UAVs. The system combines several image processing techniques, including feature
extraction, object segmentation, and optical flow analysis, to identify objects in real-time.
The system's performance is evaluated using a data-set of UAV images, and the results
show that it can achieve high accuracy in detecting and tracking objects, even in
challenging environments [7]. A method for texture classification that can be used for
object detection. The method uses local binary patterns to extract features from images at
multiple scales and orientations, which are then used to train a classifier. The paper presents
experimental results that show the effectiveness of the proposed method in detecting
textures and objects in images [8]. Object detection algorithm called faster R-CNN, that can
2
E3S Web of Conferences 391, 01016 (2023) https://doi.org/10.1051/e3sconf/202339101016
ICMED-ICMPC 2023
achieve high accuracy while maintaining real-time performance. The algorithm uses a
convolutional neural network (CNN) to extract features from images, which are then used
to generate object proposals. The object proposals are refined using a region-based CNN,
which produces a final set of object detection. The paper presents experimental results that
show the proposed algorithm outperforms previous state-of-the-art methods in terms of
accuracy and speed [9]. Another algorithm called, You Only Look Once (YOLO) Unified,
Real-Time Object Detection, which presents an object detection algorithm that can detect
objects in real-time with high accuracy. The algorithm uses a single neural network to
predict object classes and bounding boxes directly from input images, eliminating the need
for region proposals. The paper presents experimental results that demonstrate the
effectiveness of the proposed algorithm in detecting objects in various datasets [10].
OpenCV library, which is a free, open-source computer vision library designed for real-
time applications. The paper outlines the motivation for developing OpenCV, the key
features of the library, and its architecture. The author explains that computer vision has
traditionally been a complex and challenging field due to the diversity of image and video
processing applications. OpenCV was designed to provide a common platform for
computer vision research and development. The library is intended for use in real-time
applications such as robotics, surveillance, and image and video processing. The paper
describes the key features of OpenCV, including its image and video processing functions,
feature detection and extraction algorithms, and machine learning capabilities [11].
After observing the various research studies, we decided that the computer vision is the
most suitable technology for detecting the objects in our project. In conclusion, these
papers demonstrate the potential of advanced techniques and algorithms for object
dimension detection. However, there is still room for improvement, as this is a very modern
field of study. The object detection detection is also a contributor of advancement of
robotics and AI.
3 Proposed approach
In this paper, we propose an approach that uses the video or image of the surroundings
captured by the computer's webcam or external camera as an input to identify items,
measure their dimensions, and identify them.With a stand, we are able to keep the camera
at a specific distance while also adjusting the camera's height, width, and depth. With the
use of this system, we are able to identify many objects at once, provide their
dimensional measurements, and obtain other information such as the object's area of
occupancy. The system is created using a python program which makes use of python
libraries for queue, math, numpy, and computer vision (opencv-cv2). This system is
composed of three modules(a,b and c) that act in accordance with various system-related
tasks and are discussed below sequentially.
Module a - the first module is used for obtaining the video frames input, in this module
we configure our environment by setting the camera width, height and also we set the frame
rate of the video input.
Module b - the second module is used to detect object by capturing the objects
boundaries. There are multiple functions performed in this module. Initially the input frame
to gray scale to get better understanding of the details in input. Then we apply Gaussian
blurring this methodology substitutes a Gaussian kernel for a box filter. The
cv.GaussianBlur() function has completed its work.The kernel's width and height should be
positive and odd, accordingly. Furthermore, we must supply sigmaX and sigmaY, which
stand for the standard deviations in the X and Y axes, respectively. sigmaY is assumed to
be the same as sigmaX if only sigmaX is supplied. If either or both are provided as zeros,
values are chosen based on the kernel size. In order to exclude Gaussian noise from an
3
E3S Web of Conferences 391, 01016 (2023) https://doi.org/10.1051/e3sconf/202339101016
ICMED-ICMPC 2023
image, Gaussian blurring is highly effective.Then image thresholdinng cv.threshold() is
applied, the thresholding is achieved through using procedure cv.threshold. The provided
image, which must be in grayscale, is the first argument. The threshold value, which is used
to classify the pixel values, is the second input. The following input determines the highest
value that will be given to pixel values that are higher than the threshold. The function's
final parameter, which is obtained using OpenCV, supports multiple types of
thresholding. Then contours are detected three choices exist when employing the
cv.findContours() function: source picture, contour retrieval mode, and contour
approximation method. And the hierarchy and contours are produced. List of every contour
in the image is identified as contours.The (x,y) coordinates of an object's border points are
arrayed for each contour.
Module c - the third module is used to find the object dimensions by using the contour
data we find the various lengths of the object from the math library we use hypot method
which is used to calculate euclidean distance, there by allowing us to calculate
length,breadth, and also also area of the object.
Fig. 1. Block diagram of proposed approach
Fig. 2. Block diagram of working procedure
4 Results
Here we present sample outputs from our proposed approach involving various objects, and
in table 2 we determine the accuracy using the results of the object dimensions. In Fig. 3 it
4
E3S Web of Conferences 391, 01016 (2023) https://doi.org/10.1051/e3sconf/202339101016
ICMED-ICMPC 2023
is showing mobile phone is detected and screen is displaying the dimensions and it’s area.
In Fig. 4 it is showing pen is detected and screen is displaying the dimensions of pen and
it’s area.
Table 2 is showing the results for various objects we tested. In this table, we recorded
the measurements of different objects and compared with the actual dimensions of the
objects and then calculated the percentage error to get the accuracy of the measurement.
Percentage Error = |OA − AA |/|AA|
OA=Observed Area
AA=Actual Area
Accuracy(%) = 100 - Percentage Error
Fig. 3. Sample output 1
Fig. 4. Sample output 2
5
E3S Web of Conferences 391, 01016 (2023) https://doi.org/10.1051/e3sconf/202339101016
ICMED-ICMPC 2023
Table 1. Results of Proposed system being tested on various objects
Actual Actual Actual Observed Observed Observed
Object Length Breadth Area Length Breadth Area Accuracy
(mm) (mm) (mm*mm) (mm) (mm) (mm*mm) (%)
Mobile 155 72 11160 153.29 70.29 10774.47 96.54
Pen 120 13 1560 119.10 13.09 1559.19 99.95
Book 275 164 45100 276.01 163.5 45127.63 99.94
Box 302 305 92110 294.50 310.1 91324.45 99.14
Wallet 91 73 6643 92.74 74.69 6926.75 95.72
Pencil 140 45 6300 141.1 45.0 6349.5 99.21
Box
Adapter 34 47 1598 34.99 48.46 1695.61 93.89
5 Conclusion
In this paper, we have used computer vision because it helps computers to process, analyze,
and understand the digital videos. The proposed work shows the application of computer
vision that exhibits a significant level of accuracy for object detection and dimension
measurement. The suggested work is used in getting the dimensions of the objects with an
accuracy of over 95% most of the times, these dimensions are used in calculation of the
area occupied by the object. Our proposed work can find various applications, including
every day use, e-commerce applications, Industrial use and AI Automobiles , e.t.c.
References
1. G. Chandan, Mohana and A. Jain, Real-Time Object Detection and Tracking by Using
OpenCV and Deep Learning, ICIRCA, (2018).
2. W. Wang, Z. Li, J. Deng, H. Yao, X. Xuan, Z. Wang, A review of research on an object
detection based on deep learning, AINIT, (2020).
3. S.Manjula, Dr. K. Lakshmi, A study on object dectection, IJPTFI,ISSN: 0975-766X,
(2016).
4. U. Binay , N. Yadav , Comparative Study of Object Detection Algorithms, IRJET-
V4I1103, Volume 4, Issue 7, (2017).
5. Brain Team, V. Quoc Le, Google Research, R. Pang, M. Tan, EffificientDet: Scalable
and Effificient Object Detection, CPR, (2020).
6. Sunil, Gangandeep, Study of object detection methods and applications on digital
images, IJSDR1905088, Volume 4, Issue 5, (2019).
7. K. Shreyamsh, UAV: Application of object detection and tracking techniques for
unmanned aerial vehicles, Texas A & M University, (2015).
8. M. Pietikinen, T. Ojala, T. Maenpaa, Multi-resolution gray scale and rotation invariant
texture classification with local binary patterns, IEEE Transactions on Pattern Analysis and
Machine Intelligence, Volume 24, Issue 7, 971 - 987, (2002).
9. R. Girshick, Fast R-CNN, In Proceedings of the IEEE international conference on
computer vision pp. 1440-1448, (2015).
10. R. Girshick, , S. Divvala, J. Redmon and A. Farhadi, (YOLO) You only look once:
Unified, real-time object detection, arXiv preprint arXiv:1506.02640, (2015).
11. G. Bradski, OpenCV: A general purpose computer vision library, Computer Vision,
(2008).
