Method, an apparatus and a computer program product for object detection

An apparatus comprising at least one processor, memory including computer program code, the memory and the computer program code configured to, with the at least one processor, cause the apparatus to:

• receive a video comprising video frames as an input;
• generate a set of object proposals from the video;
• generate object tracklets comprising regions appearing in consecutive frames of the video, said regions corresponding to the set of object proposals with a predetermined level of confidence;
• construct a graph comprising the set of object proposals and superpixels grouped from pixels of a frame;
• calculate a first cost function for superpixel likelihoods in terms of the object proposals in an object tracklet;
• calculate a second cost function for object proposal likelihoods in terms of the superpixels in the frame;
• minimize the first and the second cost function with each other;
• compute posterior probabilities of the superpixels;
• and
• assign each superpixel with an object class with a maximum posterior probability to constitute semantic object segmentation.

A computer program product embodied on a non-transitory computer readable medium, comprising computer program code configured to, when executed on at least one processor, cause an apparatus to:

• receive a video comprising video frames as an input;
• generate a set of object proposals from the video;
• generate object tracklets comprising regions appearing in consecutive frames of the video, said regions corresponding to the set of object proposals with a predetermined level of confidence;
• construct a graph comprising the set of object proposals and superpixels grouped from pixels of a frame;
• calculate a first cost function for superpixel likelihoods in terms of the object proposals in an object tracklet;
• calculate a second cost function for object proposal likelihoods in terms of the superpixels in the frame;
• minimize the first and the second cost function with each other;
• compute posterior probabilities of the superpixels;
• and
• assign each superpixel with an object class with a maximum posterior probability to constitute semantic object segmentation.