Performing object detection in an image

A method for performing object detection in an image at a single-instruction, multiple data processor having a set of computation units arranged to execute a plurality of classifier routines, each classifier routine having identical instructions to the other classifier routines, and the method comprising:

• executing a classifier routine in parallel at each of the set of computation units so as to test for a plurality of object features in a plurality of image regions such that the classifier routines are configured to concurrently test different features against different image regions using data identifying an image region and a feature under test;
• wherein, on testing an identified feature against an identified image region, passing or failing the identified image region;
• and
• maintaining an index representing an extent of search by the set of computation units through the plurality of image regions, and each classifier routine, in dependence on whether the identified image region passes or fails, incrementing the index so as to update the progress and cause an identifier of the next image region to be returned to the classifier routine.

A non-transitory computer readable storage medium having stored thereon computer readable instructions that, when executed at a computer system comprising a single-instruction, multiple data processor, cause the computer system to perform a method for performing object detection in an image at the single-instruction, multiple data processor having a set of computation units arranged to execute a plurality of classifier routines, each classifier routine having identical instructions to the other classifier routines, the method comprising:

• executing a classifier routine in parallel at each of the set of computation units so as to test for a plurality of object features in a plurality of image regions such that the classifier routines are configured to concurrently test different features against different image regions using data identifying an image region and a feature under test,
• wherein, on testing an identified feature against an identified image region, passing or failing the identified image region;
• and
• maintaining an index representing an extent of search by the set of computation units through the plurality of image regions, and each classifier routine, in dependence on whether the identified image region passes or fails, incrementing the index so as to update the progress and cause an identifier of the next image region to be returned to the classifier routine.