Eyes、Ears、Nose, Tongues...those parts

@ the NCCU Art & Culture Center

《Eyes、Ears、Nose, Tongues...those parts》作為臉部的感官部件，展覽從顏面為人類接受訊息以及最接近大腦的人體部位作為出發，討論人類物種有著不同的臉部相貌，並延伸由視覺去辨識臉部作為個人與主體的象徵與代表，甚至為身份確認的方法。

在近代電腦視覺的演進，Facial recognition（臉部辨識）系統從基於類間變化及類內變化的方式，藉由機器學習轉為接近強人工智慧（Strong AI）的方向。同時，隱私權也因為網路與監視科技的發展變得越來越為重要。未來科技發展亦會逐漸朝向Passwordless（無密碼登入）方向演進，所以生理訊息類的加密方式與辨識將會成為研發重點。例如：臉部解鎖，這類的應用為目前個人行動裝置身份確認的方法之一。

臉，我們藉由眼睛觀看對方，直接、間接透過螢幕或其他方式認知個體。作品《這可能是你》以網路即時留言者暱稱作為搜尋臉部圖像的關鍵，藉由合成資料與機器學習，產出許多不同的亞洲臉孔以及假個資，創造出「這有可能是你」的情境假象，用以質問與顯現個人的數位足跡與龐大網路資料能夠作為軟體系統對於人類的想像。這種方式並非像透過光的折射，再藉由大腦的化學生理變化去理解畫面，而是藉由每一個Pixel dot（像素點），再經過演算方式的一連串處理，當電腦軟體將人形辨識為物件時，知道依據型態去定義人，再加上物件是否移動的邏輯，它能更確認這是個會動的人，而不是一個靜態的人型看板。軟體藉由參考行為與環境的參數及變化，增強對於物件或事件的定義。

作品《肖像2011》運用回收材料製作機械人頭，賦予機械單一種人類的情感表現方式。當展覽現場沒有人時，裝置會開始哭泣流淚。人在情感波動下，哭泣的行為表現方式，許多人會將之隱藏。這是一種安靜隱密的情緒，隱藏的行為可能較難在冷冰冰的機械上表現，但在現今網路電商的消費者行為分析，追蹤喜好的廣告投放，或許能找到相似的表現。某類商品雖然較優惠，但因為消費者不常瀏覽這類的商品，所以就不寄送電子報或是推播通知。這類追蹤消費者使用習慣的方式，由網站記錄何時瀏覽，據以判斷於何時投放促銷方案，才能促進買氣。

上述建構在基礎網路架構的資料搜集，其中以CCTV閉路與網路攝影系統最為常見，但囿於各地法律，而具不同的規範。作品《證件照拍攝指南》為2021年與陳珊妮、談宗藩合作的共同行動。該行動起源於陳珊妮遺失身分證，我們運用IoT（物聯網）搜尋引擎「Shodan」找尋在臺北市區安全性較低的網路攝影機，透過攝影機畫面判斷攝影機地點，然後到現場拍攝身份證件照。照片沖洗完成後，再到戶政事務所補辦身份證。簡單的駭客行為作為圖像的產出，將其圖像放置在國家身份證件上，是開始這個行動的出發點之一。《證件照拍攝指南On-Site》為2023年政大駐校計畫再創作版本，將延伸2021年的版本而為實體裝置作品。

The exhibition 《Eyes, Ears, Nose, Tongues...those parts》 Those Parts revolves around the face in order to discuss issues of identity related to contemporary uses of technology, as online consumer behaviour and privacy, facial recognition and surveillance systems. The face is arguably the part of the human body that is most receptive to receiving messages. It is also the closest part to the brain. In machine learning systems and network infrastructures based on data recollection, the visual recognition of faces serves as a symbolic representation of individual subjects and as a method of identity verification.

With the evolution of computer vision, in recent times facial recognition systems have evolved from being based on 'interclass' and 'intraclass' variations to being directed towards strong artificial intelligence through machine learning. At the same time, privacy has become increasingly important due to the development of the internet and surveillance technology. In the future, technology will also gradually evolve towards 'passwordless' login, so the encryption and identification of physiological information will become a focus of research and development. For example, facial unlocking is currently one of the methods most commonly used for identity verification on personal mobile devices.

Through the face we visually perceive individuals directly (through the eyes) or indirectly, ie. through other means such as screens. The work "It Could Be You" uses the online pseudonym of the real-time message sender as a key to search for facial images, creating a large number of different Asian faces and fake personal information through synthetic data and machine learning that creates the illusion of a potential identity. This is an attempt to question and reveal the digital footprint in a vast network data base as a software system that could come close to human imagination or memory recollection. Instead of receiving and understanding the image through the refraction of light and the chemical and physiological changes in the brain, the process used is based on a calculation of each pixel dot that recognises facial features and individual characteristics through a series of operations. When the computer software recognises the human form as an object, it 'learns' how to physically define a person based on their type and appearance and, with the added logic of whether or not the object is moving, it can more accurately confirm that this is a subject rather than a static mannequin. The software enhances the definition of objects or events by referencing the parameters and changes of behaviour and environment.

The work "Portrait 2011" uses recycled materials to create a mechanical human head that attempts to express human emotions. When there is no one at the exhibition space, the device will start to cry and shed tears. Many people hide their crying when they are emotionally unstable. This is a quiet and secretive emotion difficult to express on a cold and mechanical object, but in the current analysis of consumer behaviour in online shopping, ie. tracking the preferences of advertisements, it may be possible to find similar expressions. Even when a certain type of product is more favorable, if consumers do not often browse this type of product, they will not receive newsletters or push notifications. This method of tracking consumer usage habits can be used to determine when to promote sales in order to boost them.

The construction of the aforementioned works is based on the data collection performed by basic network infrastructures. Another way to approach this can be found in CCTV closed-circuit and network photography systems, which in addition are subject to different regulations due to local laws. "ID Photo Shooting Guide" is a joint action in 2021 with Sandee CHAN and TAN Tsung Fan. The action originated when Sandee CHAN lost her ID card. Then, we used the IoT search engine "Shodan" to find low-security network cameras in Taipei city and determine their location to take ID photos. After the photos were processed, we went to the household affairs office to reissue the ID card. Simple hacking behaviour as the production of images, placing the image on a national identity card, is one of the starting points for this action. "ID Photo Shooting Guide On-Site" is a re-creation of the 2023 political university residence project that extends the 2021 version to be a physical device work.