Eating is an essential part of our daily lives. Detection of accurate food intake leads to automatic dietary assessment. Conventional chewing (mastication) detection methods primarily use electromyography which requires attaching and detaching myoelectric pads. In this study, we developed a wearable chewing detection device that uses a pressure sensor array and a shock-absorbing material integrated into a temple frame of eyeglasses. The proposed design eliminates the need for adjusting the sensor placement in advance, as the developed eyeglass-type device ensures consistent contact with the wearer’s temporal region. The deformation of the temporal region during chewing is captured by the pressure sensor array as the pressure changes. We analyzed the data when wearers chewed food samples and proposed an algorithm to detect chewing by using a dynamic adaptive threshold. The fundamental experiment showed that the algorithm can detect the timing of chewing with a maximum delay of 58 ms from the detection time point to the sound peaks of the food breakdown sound as a reference.