SS06a-21 Smart Inspection System for Construction

Subscribe
Notify of
7 Comments
Inline Feedbacks
View all comments
SHEN, Shanpu(@eeshanpu)
January 19, 2022 3:48 pm

An interesting work. How accurate is the method for image classification? Is that possible to have some errors? How to make it robust and reliable?

CHEN, Sida(@schenci)
January 19, 2022 3:54 pm
Reply to  SHEN, Shanpu

We are at the stage of sticker recognition, so we haven’t tested image classification yet. It is always possible to have some errors for a network. We planned to add more critical information on sticker to aid classification.

CAO, Xuanyu(@eexcao)
January 19, 2022 3:27 pm

Interesting. What is the main technical challenge of implementing this system? Is it the image processing part?

CHEN, Sida(@schenci)
January 19, 2022 3:51 pm
Reply to  CAO, Xuanyu

Indeed. Also the integration of different hardware modules is quite challenging as well.

TANG, Erqiang(@etangaa)
January 19, 2022 3:01 pm

A very interesting project! I’d like to know if you plan to implement the image classification part? And how would you like to separate the sticker from the background?

CHEN, Sida(@schenci)
January 19, 2022 3:15 pm
Reply to  TANG, Erqiang

Thank you for your interest! We will try to find a suitable network such as YOLO to do the classification. Separating sticker from background can be done with 2D code because we can redesign the sticker.

TANG, Erqiang(@etangaa)
January 19, 2022 3:16 pm
Reply to  CHEN, Sida

That’s great. Thanks for your reply!