Evaluating high-resolution precipitation forecasts is challenging because of the demonstrated inappropriateness of traditional grid-point based verification metrics. One solution to this challenge is an object based approach. Such an approach is adopted in this study to diagnose and verify the characteristics of precipitation forecasts during the 2009 NOAA Hazardous Weather Testbed Spring Experiment. The distributions of various attributes of the forecasted objects from all ensemble members, members using the ARW model and members using the NMM model are compared to those of observed objects. Object based forecast accuracy is also quantified using two measures of the similarity of the forecast and observed fields, the Object-based Threat Score and the Median of Maximum Interest, for each group of ensemble members. In general, forecast objects are more numerous than observed objects with a smaller median area, more circular median aspect ratio, and more eastward median centroid location than the observed objects. There is a more skewed distribution of east-west centroid location, a flatter distribution of orientation angle, and less variable area distribution for forecast objects than observed objects. Members using the ARW model have fewer objects with a more northward median centroid location after the 12 hour lead time, a more linear median aspect ratio and smaller median area than members using the NMM model. The group of ARW members has higher accuracy and is more likely to contain the most accurate member than the group of NMM members. After applying an object based bias adjustment, the differences in accuracy between the ARW and NMM members are reduced.