Harmful algal blooms (HABs) are associated with water quality degradation, which damages fisheries, ecosystems, and public health. Consequently, techniques for detecting the occurrence of HABs have been developed globally. However, conventional techniques are unable to discriminate the causative species/genera of HABs objectively. Therefore, the aim of the present study was to develop a technique for frequently analyzing the causative genera of HAB events using repeat digital photography with a stationary device named “HABcam.” The HABcam was installed at Okigamisenishi monitoring station in the inner western area of Ariake Bay, Japan, and the digital images obtained were analyzed to quantify the sea surface color and to estimate the probability of a HAB occurrence on a daily basis using Bayes' theorem. The estimated probability of a HAB occurrence accurately discriminated a HAB occurrence or non-occurrence in the research area on 19/21 days (=90.5%) for Chattonella spp. and 7/8 days (=87.5%) for Skeletonema spp. These findings indicate that this technique can be used to objectively determine the causative genera during HAB events and to observe HABs with high accuracy and at a high frequency.