<p> A&#xa0;dual-emission ratiometric fluorescence probe was constructed for detection of perfluorooctanoic acid (PFOA) by hybridizing MoB quantum dots (MoB QDs) and nitrogen, fluorine co-doped carbon dots (N, F-CDs). As the concentration of PFOA increased, the fluorescence emission of MoB QDs/N, F-CDs at 560&#xa0;nm was quenched due to fluorine-fluorine interaction, while the fluorescence emission at 440&#xa0;nm kept unchanged, resulting in the fluorescence color changing from yellow to green. A good linear relationship was observed between <i>I</i>₄₄₀/<i>I</i>₅₆₀ and the concentration of PFOA in the range 0.01–90&#xa0;μM and 90–450&#xa0;μM, with a detection limit as low as 0.01&#xa0;μM. The test strips for PFOA were prepared for on-site visualization and quantitative determination. The fluorescence images of the test strips were acquired using the&#xa0;camera in the smartphone, and the signals were converted into RGB values by color recognition software. The obtained <i>G</i>/<i>R</i> value had a good linear relationship with the concentration of PFOA ranging from 0.5 to 200&#xa0;μM. The deep learning model of YOLOv5 algorithm was utilized for data analysis to make the fluorescence detection more accurate. Therefore, the ratiometric fluorescence sensing system is established with easy operation, short response time, and low cost for on-site detection of PFOA in real water samples and textiles, which is of practical significance for environmental water monitoring and human health.</p> Graphical Abstract <p></p>