Robust Pooling Through the Data Mode

The task of learning from point cloud data is always challenging due to the often occurrence noise and outliers in data. Such inaccuracies can significantly influence performance state-of-the-art deep networks their ability classify or segment objects. While there are some robust approaches, they computationally too expensive for real-time applications. This paper proposes a solution that includes novel pooling layer which greatly enhances network robustness performs faster than approaches. proposed looks mode/cluster using two methods, RANSAC, histogram, as clusters indicative models. We tested into frameworks such Point-based graph-based neural networks, tests showed enhanced compared methods.