This research presents preliminary results of developing aerial imaging platform for disaster site surveillance. The Can-Satellite (CanSat) less than a 700g weight equipped with an RGB camera and a forward looking infrared (FLIR) camera which can be launched with the rockets, drones, or balloons was developed. Additional flight sensors and network interfaces were added for the data transmission and coordinate recognition of the detected survivor. The deep neural network architecture was used to classify the human objects in the transmitted images. We expect that the developed platform can be rapidly launched to the disaster site under the bad weather conditions and multiple CanSats can be used simultaneously without concerning the loss due to low cost to build a single device.