Prior ANN-SNN conversion works concentrated on frame-based dataset of classification and detection, etc., whereas in this work, we demonstrate a converted SNN for image segmentation that is applied to a natural video dataset. Layer-specific normalization is applied to replace the uniform normalization in an effort to obtain low latency and high accuracy in SNN networks. We also present interval reset with early stopping to reduce the latency of SNNs, and evaluate the resulting accuracy-latency tradeoff. We find that when s=20, alpha=0.2 and Tp=3, the simulation timesteps can be reduced by 35.7x with 1.5% accuracy loss.