What is Computer Vision?
If I ask you what is in the photo below, what do you see? You’ll probably answer the following: trees, boats, beach, people, rocks, cloud and so on. This is a very easy task for a person that even an average 5-year-old kid can identify. Computer Vision or often called as “CV” is a field of study that assists computers to “see” and “identify” things within photos and videos [1]. A person can identify the content of a photo easily. A person can simply describe the content of a video without a problem. This is exactly what computer vision wants to achieve, to imitate human vision.
Applications of Computer Vision
Where do we use computer vision, you might ask? In today’s world, technology is constantly evolving in a fast rate. To name a few, computer vision is used in the following:
Face recognition: this could be seen in social media such as Snapshot, Messenger, TikTok and Instagram. Computer vision is used to identify your face and apply those cute or scary filters. It identifies where your eyes are, your nose, lips, eyebrows and every little detail. It is also used to unlock some phones and laptops.
Content-Moderation — Social media platforms such as Facebook uses CV to moderate the feeds. For example, if someone posted a video or photos that are not appealing to the eye (for example: violence and extremism), the media is not shown right away or perhaps removed or deleted. Facebook blocks it and if the post is not deleted, it requires you to click a button to see the content. CV is used to distinguished these kinds of posts.
Self-driving cars: how do you think Tesla’s car drive itself? You guess it right, computer vision plays a huge role to detect lanes, traffic lights, signs, cars and others. Although it is much more complex than that, as it involves Artificial Intelligence (AI). Computer vision is merely like the “eye” field of AI.
Gaming [3]: if you have played Microsoft Kinect, you would know how fun it is to play this game which requires you to move or dance in front of the camera to play their games. Yes, it uses sensors, but it also uses computer vision to detect a player. The camera sees where you are and you must be located at a certain distance from the camera. It detects and sees your movements and sees how many players have joined.
Surveillance and Shopping: Shopping surveillance can be used to track how many people come and go. One good example, is Amazon Go and Amazon Go Grocery which is basically a shopping store with variety of options such as snacks, bakeries, lunch, fresh produce, and even household items [4]. The idea is you scan your QR code, shop and grab items from the shelf and go. No waiting in line for cashiers, and you get charged through your app. Isn’t that amazing? I have been to one of their stores in California and there were multiple of cameras around the store. Computer vision, along with sensor fusion and deep learning, tracks which products are taken out of the shelf or placed back to the shelf. The camera sees what you take, and this is one of the uses of CV.
Limitations of CV
CV combined with deep learning and AI has been vastly developing in today’s world. CV has enabled detection of objects in photos and videos but are only limited to what it has been trained to do. Computer Vision is merely analyzing and performing algorithm through media’s pixels but has no understanding of what’s going on, unlike human visions. There is much more development that is needed.
Stay tuned for the next blog, as I will be expanding more on why to use CV and how CV works.
References:
[1] B. Dickson, “What is computer vision?,” 14 Jan 2019. [Online]. Available: https://bdtechtalks.com/2019/01/14/what-is-computer-vision/. [Accessed 27 Jan 2021].
[2]M. A. Rafael, “El Nido, Palawan PH,” 11 Jan 2019. [JPEG File]. [Accessed 27 Jan 2021].
[3] J. Le, “The 5 Computer Vision Techniques That Will Change How You See The World,” 12 Apr 2018. [Online]. Available: https://heartbeat.fritz.ai/the-5-computer-vision-techniques-that-will-change-how-you-see-the-world-1ee19334354b. [Accessed 27 Jan 2021].
[4] Amazon.com, “Amazon Go,” 2021. [Online]. Available: https://www.amazon.com/b/ref=s9_acss_bw_cg_agogo_2b1_w?node=20931384011&pf_rd_m=ATVPDKIKX0DER&pf_rd_s=merchandised-search-20&pf_rd_r=H2GTJ3W0Z2ZPNADRAMY3&pf_rd_t=101&pf_rd_p=d510bfd7-cb66-42fa-820f-51e50a76df14&pf_rd_i=16008589011. [Accessed 27 Jan 2021].