The acoustic and seismic sensor system, with its advantages of low cost, low power consumption, ease of deployment, strong obstruction resistance, and high concealability, is particularly well-suited for the development of environmental activity recognition systems. However, existing classification methods for acoustic and seismic signals are generally constrained to closed-set scenarios, where the test-phase input classes must align with the training-phase classes. This limitation significantly hinders the deployment of such sensor systems. To overcome this challenge, open-set recognition is essential. In this chapter, we propose an open-set recognition algorithm, Cross-layer Selected Class Anchor Clustering Learning, which leverages selective learning based on CAC Loss and incorporates a cross-entropy constraint in the penultimate layer of the network. To evaluate the performance of our algorithm, we have constructed a comprehensive acoustic and seismic dataset that includes typical environmental activities, and compared our method with other state-of-the-art algorithms. The results demonstrate that the proposed algorithm not only achieves effective classification of closed-set samples but also delivers high recognition accuracy for open-set unknown samples.