Objective: Solves the problem that the off-line detection method is adopted in the current cut tobacco structure detection, and the detection result lags behind. Methods: Using the air separation dilution method and image intelligent recognition technology, the cut tobacco structure was detected in the air separation room of the silk production line, the length, width and area of cut tobacco were detected online, and the proportion of whole and broken cut tobacco was counted. Results: By comparing the existing off-line detection instruments for cut tobacco structure. The detection device had high measurement accuracy and good stability, and could truly reflect the actual quality of cut tobacco structure of silk production line. Conclusion: The results show that the device can effectively reduce the workload of inspectors.