1 | initial version |
yes, it is so. (basically it was trained on the inria dataset)
you can also try the upperbody haar cascade, or the dpm module from opencv_contrib
2 | No.2 Revision |
yes, it is so. (basically it was trained on the inria dataset)dataset, which only has upright people)
you can also try the upperbody haar cascade, cascade(for sitting), or the dpm module from opencv_contrib
trying to find persons lying on the ground is extra hard !
3 | No.3 Revision |
yes, it is so. (basically it was trained on the inria dataset, which only has upright people)
you can also try the upperbody haar cascade(for sitting), or the dpm module from opencv_contrib
trying to find persons lying on the ground is extra hard !! (even if you try to rotate the image 90°) there's not much left of usual human features, shadows, etc.
4 | No.4 Revision |
yes, it is so. (basically it was trained on the inria dataset, which only has upright people)people, also mostly front & back, not many profile samples)
you can also try the upperbody haar cascade(for sitting), or the dpm module from opencv_contrib
trying to find persons lying on the ground is extra hard ! (even if you try to rotate the image 90°) there's not much left of usual human features, shadows, etc.