Anthracnose is one of the most common diseases of camellia oleifera, so it is of great significance to quickly detect it. In this paper, a new method for the detection of anthracnose in camellia anthracnose leaves is proposed. First, laser induced breakdown spectroscopy is used to quickly and non-destructively diagnose the Mn element content in the leaves of normal and infected anthracnose camellia oleifera leaves. The true content of Mn element in the samples is analyzed by flame atomic absorption spectrometry. Second, different pretreatment methods, such as smoothing, denoising, normalization, baseline correction, first-order derivation noise reduction, and second-order derivation noise reduction, are used to preprocess the spectral data. Partial least squares (PLS) method is used to establish a quantitative model. Interval partial least-squares regression (iPLS) method is utilized to filter the spectral data. Finally, combined with 7-point smoothing and first derivative noise reduction, the quantitative model is established based on iPLS. The results show that the modeling effect of the sixth subinterval is the best when it is divided into 24 subintervals, the modeling correlation coefficient is 0.9076, the modeling root mean square error is 0.2090 μg/mg, the prediction correlation coefficient is 0.8947, and the prediction root mean square error is 0.2100 μg/mg.