An endoscopic image processing device includes one or more processors. The processors perform: detecting a region of interest for an observation image, the observation image being obtained by picking up an image of an object and being sequentially input to the processors, judging whether detection of the region of interest continues or not to acquire a judgment result, emphasizing with adding visual information for emphasizing a position of the region of interest to the observation image during a period during which the region of interest is detected. The processors cause an appearance of the visual information added to the observation image to differ between a case where the detection of the region of interest continues for a predetermined time period or more and a case where the detection of the region of interest discontinues in a time period shorter than the predetermined time period, based on the judgment result.
