Rapidly increasing car traffic creates a need for low-cost traffic monitoring solutions. The suitability of commodity wireless chipsets for traffic sensing purposes was already described in the literature: The signal propagation paths inside a wireless sensor network nearby a street are affected by the presence of a passing car. This leads to detectable field strength variations and enables traffic monitoring applications. However, all existing low-cost approaches suffer from missing spatial information, which leads to counting inaccuracies for high-density traffic scenarios. It was recently shown that the frequency dependence of the Received Signal Strength contains coarse information about the surrounding. We show that this information is suitable for traffic monitoring purposes and propose a low-cost traffic counting approach that allows to observe the traffic on a two-lane road individually for each lane. It is further shown that the counting accuracy can be significantly improved compared to existing low-cost approaches that cannot distinguish between individual lanes. Using a Bluetooth implementation, a detection rate of 99.6% with a false-positive rate of 1.1% was achieved.