Solidification cracking is one of the most critical weld defects in laser welding of Al 6000 alloys. It occurs at the end stage of solidification due to the shrinkage of weld metal and deteriorates the joint strength and integrity. Filler metal can control the chemical composition of weld metal, which mitigates the solidification cracking; however, the chemical composition is difficult to control in autogenous laser welding. Temporal/spatial laser beam modulations have been introduced to control the solidification cracking in autogenous laser welding because the welds morphology is one of the factors that influence the solidification cracking initiation and propagation. Solidification cracks generate thermal discontinuity and visual flaws on the bead surface. In this study, a high-speed infrared (IR) camera and a coaxial charge-coupled device (CCD) camera with an auxiliary illumination laser (808 nm) were employed to identify solidification cracking during laser welding. Deep learning models were developed using two sensor images of the solidified bead. The output of deep learning models were location-wise cracking formation. Convolution neural network (CNN)-based deep learning models showed excellent cracking identification accuracies. In this study, the multiple sensor system of high-speed IR camera and low-cost CCD camera demonstrated its capability and potential in laser welding inspection.