Methods and systems for generating graph representations of a scene
A computer-implemented system and method of generating a graph representation of a scene comprising receiving sensor data representative of a perceived scene captured with a sensor; defining a plurality of nodes based on the received sensor data; creating a densely connected graph by connecting each node to a number of nearest neighbour nodes; predicting, for each pair of connected nodes of the densely connected graph, at least a node probability, wherein the node probability represents a probability that the pair of connected nodes represent the same object; and generating a graph representation of the perceived scene based at least on the densely connected graph, and the predicted node probability. The method may also be applied to two or more observations of a scene.
1 . A computer-implemented method of generating a graph representation of a scene based on a single observation of the scene, said graph representation comprising a plurality of nodes and edges, wherein a node represents an object within the scene, and wherein an edge connects pairs of nodes of the plurality of nodes and represents a spatial relationship between objects within the scene, the method comprising:
receiving sensor data representative of a perceived scene captured with a sensor;
defining a plurality of nodes based on the received sensor data;
creating a densely connected graph by connecting each node to a number of nearest neighbour nodes;
predicting, for each pair of connected nodes of the densely connected graph, at least a node probability, wherein the node probability represents a probability that the pair of connected nodes represent the same object; and
generating a graph representation of the perceived scene based at least on the densely connected graph, and the predicted node probability.
2 . The computer-implemented method of claim 1 , wherein defining a plurality of nodes based on the captured perceived scene comprises:
identifying one or more objects within the perceived scene based on the received sensor data;
determining whether each identified object is an object of interest, wherein an object of interest is preferably a static object; and
defining a plurality of nodes, wherein each node corresponds to an object of interest within the perceived scene.
3 . The computer-implemented method of claim 1 , further comprising merging the pair of connected nodes when the node probability is above a first threshold.
4 . The computer-implemented method of claim 1 , further comprising removing an edge from between the pair of nodes when an edge probability of the edge between the pair of nodes is below a second threshold.