Face Detection
Face recognition systems are being „appreciated as one of the most promising applications in the field of image analysis“ (Dwivedi, 2018) and there has been a lot of research in the area of image processing in the recent years. A high quality face recognition system is carried out by „various types of algorithms used for extracting the features, their classification and matching“ (El-Sayed and Hamed, 2015) and also includes elements of „similarity measure or distance measure“. If this similarity measure in the matching parts is efficient, „the existing feature extraction method itself … can be improved“. Factors such as „pose, expression, position and orientation, skin colour, the presence of glasses or facial hair, differences in camera gain, lighting conditions, and image resolution“ (Dwivedi, 2018) are always variable and therefore complicate the method of face detection in images. The Real-Time Face Detector which „is able of detecting the faces in real-time with high accuracy“ was a real revolution that finally led to more accurate results. Face Detection „is a part of object detection“ and „the first essential step for face recognition“. There is an official classification of face detection methods which are divided into four different categories:
- Knowledge Based
- Feature Based
- Template Matching
- Appearance Based
How the Face Detection Works
The following steps are used for Face Detection on Open CV. „Firstly the image is imported by providing the location of the image. Then the picture is transformed from RGB to Grayscale because it is easy to detect faces in the grayscale.“ The following steps are image manipulation and segmentation so that the classifier can detect facial shapes more easily. In the last step the „Haar-Like features algorithm“ is used for finding the location of the human faces in a frame or image. All of the human faces have some specific features in common, for example „the eyes region is darker than its neighbour pixels and nose region is brighter than eye region“. Generally the haar-like algorithm is used to select the essential features and then extract them for face detection.
After the definition of the region of interest by assigning the coordinates x, y, w, h, the face will be detected within the frame of a rectangular box. Some additional detection techniques in this process are for example „smile detection, eye detection, blink detection“.
Potential and Challenges of Face Recognition
On the one hand face recognition can have many advantages and lead to a simplification of many processes. Because you are being identified automatically, there is no additional effort of doing a log in with data or passwords to activate a specific process. But on the other hand face recognition might make the protection of those processes less safe due to potential errors. At the same time there is the risk of potential abuses and question concerning privacy as well as the protection of individuality. On the one hand face detection might be helpful to find and identify dangerous criminals more easily and therefore contribute to more safety. At the same time it might lead to more comfort and efficiency by simplifying specific operations. But on the other hand scenarios such as the social credit system in China where you are being identified and judged on your actions at all times also remind us of the potential dangers those systems might have for us in the future.
Humanisation in the Current Design World
Humanisation is either defined as „the process of making something less unpleasant and more suitable for people“ (Cambridge Dictionary, 2022) or „the process of making something that is not human seem like a person, or treating something that is not human as if it is a person“. Both defined processes are a part of the daily work of a designer, in the second case facial shapes – or the resemblance of facial shapes – becomes crucial. In my first blog post I have already mentioned pareidolia, which can be defined as „the tendency to perceive a specific, often meaningful image in a random or ambiguous visual pattern“ (Marriam Webster, 2023). One example of the phenomenon of pareidolia in the history of art might be Giuseppe Arcimboldo who „painted collections of fruits, vegetables and other objects to look like human portraits“. But also the famous artist Leonardo da Vinci has already noticed the „tendency to perceive a meaningful image in a random pattern“.
Just like Archimboldo who saw facial elements within vegetable configurations and depict them afterwards, humans of our recent times are able to perceive face-like configurations in everyday products. Consumer behavior can be influenced by the „perception of pareidolic emotion“ (Noble, 2023) which means that people are able to „perceive core human emotions in products with pareidolic configurations“. Consumer metrics such as attention capture or the likelihood to purchase are being affected, still there is a „variation in percieved emotional intensity“ depending on the features of a specific subject and the context it has been set into. „Products with ‘happy’, ‘angry’ and ‘surprise’ configurations were likely to capture attention/promote product exploration, but only ‘happy’ products retained this advantage for purchasing decisions.“ Because the perception and engagement into processes is always depending on differences within individuals and the specific context, creatives need to be aware of the effect of pareidolia before making crucial design decision. When used efficiently pareidolia can be very beneficial and lead to an improved user experience, more personal attachment to a product or service, higher enthusiasm and greater levels of understandability and emotional involvement.
… if you look at any walls spotted with various stains or with a mixture of different kinds of stones, if you are about to invent some scene you will be able to see in it a resemblance to various different landscapes adorned with mountains, rivers, rocks, trees, wide valleys, and various groups of hills.
Leonardo da Vinci
Sources:
Yang, M.-H., Kriegman, D.J. and Ahuja, N. (2002) Detecting Faces in Images: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 24.
http://dx.doi.org/10.1109/34.982883
„Face Detection for Beginners“. Medium. Divyansh Dwivedi. April, 2018.
https://towardsdatascience.com/face-detection-for-beginners-e58e8f21aad9
Journal of Software Engineering and Applications. Vol.8 No. 9, September 2015.
https://www.scirp.org/journal/home?journalid=45
„Meaning of humanization in English. Cambridge Dictionary.
https://dictionary.cambridge.org/us/dictionary/english/humanization
Pareidolia. Noun. Mariam Webster. Est. 1828.
https://www.merriam-webster.com/dictionary/pareidolia#:~:text=-%CB%88d%C5%8Dl-y%C9%99%20%3A%20the%20tendency%20to%20perceive%20a%20specific%2C,see%20shapes%20or%20make%20pictures%20out%20of%20randomness.
„Face pareidolia in products: the effect of emotional content on attention capture, eagerness to explore, and likelihood to purchase“. Willey. Online Library. Applied Cognitive Psychology. Erin Noble, July 2023.
https://onlinelibrary.wiley.com/doi/10.1002/acp.4105
