Abstract: To automatically obtain information (number, gender, and body temperature) of indoor people effectively, adjust and control indoor equipment, and improve the comfort of indoor personnel, an information detection system of people for indoor environment control based on infrared thermal image and visible image fusion was proposed in this paper. First, visible light images and infrared thermal images in the indoor scene was acquired synchronously, and images were matched by useing camera's field of view angle. To tackle the problem that the traditional face-based person detection method was easily affected by facial occlusion and head multi-pose, which caused the decline in detection accuracy, the faster region-based convolutional neural network (Faster R-CNN) head detection algorithm was used to accurately detect and count people in visible light images. According to the head region, a gender detection algorithm based on deep learning was used to determine the gender of the person. Further more, human head regions captured in the visible light image were mapped to corresponding positions of infrared thermal images. Using the temperature scale of infrared thermal image and the color mapping, the body temperature of the indoor personnel was automatically calculated to improve the independence and operability of the system. The experimental results show that the system can automatically detect people and obtain the number, gender and body temperature of the people. At the same time, the above mentioned information can be effectively statistically and visually displayed, which provides a technical basis and data support for controlling the indoor environment.