Fine-grained place logging with Wi-Fi beacon signatures provides a useful tool for delivering various semantic location-aware services such as reminders and advertisements. Existing solutions however heavily rely on energy-hungry periodic Wi-Fi scanning for place detection in resource limited mobile devices. In this paper, we present PlaceWalker, a scheme that uses a low-power duty-cycled accelerometer in the background to continuously monitor user's significant physical activity changes (e.g., walking to resting) as it provides a useful clue to the change of place. Unlike existing schemes, PlaceWalker triggers Wi-Fi scanning only when such an activity shift is detected and then determines a change of place by comparing Wi-Fi signatures. Our experimental results verify that detecting significant activity intensity changes can precisely capture arrival/departure times, and PlaceWalker substantially lowers the energy consumption by as much as 60.9%, when compared with the state-of-the-art method. We also analyze the experimental results with a simple analytic model and validate its efficiency under varying parameter settings.