Although researchers are actively investigating methods to improve fire detector performance, few studies have investigated fire detectors that detect the type of fire. Fire type detection serves a key role in quickly extinguishing fires and preventing their spread. We present a non-dispersive infrared (NDIR)-based dual-channel mid-infrared (mid-IR) method that can detect and classify aerosol particles and gases. 4.2 μm and 4.7 μm mid-IR light emitting diodes (LEDs) light sources with strong absorption for CO2 and CO are employed. and, and the mid-IR LEDs are modulated with 900 Hz and 1,000 Hz, respectively to increase the signal-to-noise ratio and reduce interference between the light sources. The modulated lights pass through the lenses and sample, and are acquired by a photodetector. The transmittances of the 4.2 μm and 4.7 μm lights are measured to detect the aerosol particles and gases, and the aerosol particles and gases are classified via hierarchical clustering using the measured transmittances and the ratio between the measured transmittances. Various aerosol particles and gases are detected by measuring the transmittance, and the aerosol particles and gases are classified by calculating the distance between clusters. Spectral transmittances analysis of different wavelength bands will enable the detection of various aerosol particles and gases, and further improve the classification accuracy. Furthermore, this method can be applied to fire detection to develop a highly useful technique that can detect and classify fire smoke and rapidly detect the type of fire.