Detecting Number of Drift Objects in Packed Surroundings Using Coherent Motion Region

The main approach of an object detection system is to find all the possible coherent motion regions to track the moving objects which takes the locations of low-level tracked feature points as input, and produces a set of independent coherent motion regions as output. In past approaches only pedestrian is detected using a source of information in a single detector but in this, the system finds the regions based on the tracked set of features using a likelihood functions which is parameterized on locations of potential individual. For the identified coherent motion region assign a point track to at most one region to extract the subset that maximises an overall likelihood function. In case of multi-object motion, many possible coherent motion regions can be constructed around the set of all feature point tracks. The approach is robust to partial occlusion, shadows, clutter, and can operate over a large range of challenging view angles. This approach gives semantically correct detections and count of similar objects moving through crowded scenes from selected coherent regions using greedy algorithm.