AdaBoost-based face detection algorithm plays a crucial role in the field of face detection, which is based on the optimization of boosting. And its advantage is that it can determine the sample weight in the next training according to the previous sample training results. In such an algorithm, the leading research focuses on the Haar-like feature, and the algorithm is also closely related to the fundamental diagram. By extracting the portrait features using Haar-like features, calculating the eigenvalues using the Integral Image, and filtering out the weak classifier and strong classifier through the weighted voting method, the algorithm ideally improves the detection speed and accuracy of the classifier. Therefore, it has several excellent advantages, including a lower rate of generalization errors, no need for parameter adjustment, and higher stability.