Infrared (IR) cameras have versatile applications; however, the low performance and high cost of conventional photodetectors have prevented their widespread utilization in various fields. Nano-materials have recently emerged as possible sensing elements of nano-photodetectors, and have exhibited data that may outperform their conventional counterparts. Carbon nanotube (CNT), a promising nano-material with excellent optical properties, has been employed to develop high performance photodetectors with low noise, tunable bandgap, and noncryogenic cooled operation. However, it is difficult to fabricate a large scale CNT photodetector array that can be integrated into traditional IR camera structures. In order to take advantage of the outstanding performance of the nano-photodetectors and overcome the fabrication difficulties to achieve high resolution and efficient imaging, we present a compressive sensing-based IR camera using a single pixel CNT photodetector. A photonic cavity is developed to integrate with the CNT photodetector so as to increase the absorption area of the device. The camera system uses the CNT photodetector to compressively sample the linear projections of the images onto binary random patterns. Employing the compressive sensing algorithm, high resolution imaging can be achieved with many fewer samples than the original image dimension. The camera is demonstrated effectively in order to observe the dynamic movement of a laser spot. By adaptively adjusting the camera setup, a zooming technique is developed to image small features. To our knowledge, this is the first demonstration of an IR camera using a nano-size photodetector. Our work shows that compressive sensing-based cameras have the potential to complement or selectively replace conventional IR imaging systems.