The present invention addresses the issue of image analysis technology being unable to determine whether or not a dealer has accurately performed redemption, during redemption of substitute currency for gaming (chips) corresponding to wagered chips, and the issue of being unable to sufficiently prevent mistakes, as a result of the design of the chips on a gaming table being complicated. This detection system comprises a control device (9) that detects unfair practices at a gaming table (1), using image analysis results from an image analysis device (6). A substitute currency for gaming, used in this detection system, has a multi-layer structure having laminated therein a plurality of different colored plastic layers, comprising a colored layer in at least the middle thereof, and having laminated therein, on both sides of the middle colored layer, a white layer or a light-color layer (not shown in the drawing but can be a layer of any color lighter than the colored layer). In this way, the substitute currency for gaming has a multi-layer structure comprising a colored layer and has laminated on both sides of this middle colored layer a white layer or a light-color layer (not shown in the drawing but can be a layer of any color lighter than the colored layer). The camera (4) also captures an image of a plurality of stacks of game tokens placed in a plurality of bet areas set on a game table from above at an angle so that the plurality of stacks are shown in the depth direction. Also, the management control device (9) is equipped with a function of performing a deep learning image recognition to the image to determine the number or type of the gaming token in the plurality of stacks.