795 Commits

Author	SHA1	Message	Date
marina.kolpakova	8acfbde68e	remove debug detect at scale method	2012-11-14 14:22:12 +04:00
marina.kolpakova	72e2b8b370	remove size constraints of input frame	2012-11-14 14:00:02 +04:00
marina.kolpakova	08910e81af	integrate pre-Kepler architectures	2012-11-14 12:40:44 +04:00
marina.kolpakova	aa92be34d6	GK107 Policy	2012-11-12 16:37:18 +04:00
marina.kolpakova	580d8173e5	refactor computing of scaling factor	2012-11-12 11:54:40 +04:00
marina.kolpakova	0cbf9eb22a	add support for CUDA streams	2012-11-10 05:13:26 +04:00
marina.kolpakova	40600fa504	GPU version becomes algorithm	2012-11-10 05:13:19 +04:00
marina.kolpakova	e6eb1b99e1	fix negative confidence bug	2012-11-10 05:13:01 +04:00
marina.kolpakova	27eb2e27db	enable fast integral for Kepler	2012-11-10 05:12:48 +04:00
marina.kolpakova	9b251f8130	remove Sobel normalization	2012-11-10 05:12:38 +04:00
marina.kolpakova	df392cc830	use fast integral for soft cascade	2012-11-10 05:12:27 +04:00
marina.kolpakova	2bd35c4358	add support for precomputed integrals	2012-11-10 05:11:48 +04:00
marina.kolpakova	ca81628a9a	fix retrieval of detections count	2012-11-10 05:11:21 +04:00
marina.kolpakova	022a8b9698	fix rounding bug in Level creation	2012-11-10 05:11:11 +04:00
marina.kolpakova	f196e9fda4	add factory method for Fields structure	2012-11-10 05:10:42 +04:00
marina.kolpakova	0898c3c651	kernel policy	2012-11-10 05:10:35 +04:00
marina.kolpakova	916ba4c0ea	refactor preprocessing	2012-11-10 05:10:28 +04:00
marina.kolpakova	fa62e2b72f	move preprocessing into separate function	2012-11-10 05:10:13 +04:00
marina.kolpakova	ef431f70b6	fix buggy threshold zeroing if feature has zero area	2012-11-10 05:09:45 +04:00
marina.kolpakova	0424e2c8d2	remove debug code	2012-11-10 05:09:34 +04:00
marina.kolpakova	1b9bccb856	move Level into shared memory	2012-11-10 05:09:15 +04:00
marina.kolpakova	30bce16ad6	optimize roi loads ... only one thread load roi for all block	2012-11-10 05:08:56 +04:00
marina.kolpakova	fdef0adf95	Corrects objects matrix in case of the absence of objects	2012-11-10 05:08:43 +04:00
marina.kolpakova	eb91593c08	add roi support	2012-11-10 05:08:03 +04:00
marina.kolpakova	64d6e6a48d	add getROISize	2012-11-10 05:07:55 +04:00
marina.kolpakova	672cf1f445	implement different behaviour for up- and down-scaling	2012-11-10 05:07:40 +04:00
marina.kolpakova	56c7ef06e7	integrate Kepler version	2012-11-10 05:07:26 +04:00
marina.kolpakova	b52fea7fae	update soft cascade interface: - add class Detection in interface, - split sync- and async- versions, - add support for detecting at the specific scale.	2012-11-10 05:06:54 +04:00
marina.kolpakova	612a258506	kepler specific version	2012-11-10 05:06:44 +04:00
marina.kolpakova	0314e0e5d7	add kind in detection representation	2012-11-10 05:06:30 +04:00
marina.kolpakova	72b499df00	add detection storing	2012-11-10 05:06:05 +04:00
marina.kolpakova	8108bd30fe	optimize memory usage	2012-11-10 05:05:43 +04:00
marina.kolpakova	b83d4add2e	memory optimization	2012-11-10 05:05:25 +04:00
marina.kolpakova	4d9c7c1012	preprocessing ~1.981 ms	2012-11-10 05:05:17 +04:00
marina.kolpakova	1917366528	empty cascade	2012-11-10 05:05:09 +04:00
marina.kolpakova	4881205bae	refactor logs	2012-11-10 05:04:52 +04:00
marina.kolpakova	dca27b4622	fix cast bug; add logging	2012-11-10 05:04:36 +04:00
marina.kolpakova	e606a0d651	remove dead code	2012-11-10 05:04:20 +04:00
marina.kolpakova	ba50d19341	first version of soft cascade on GPU	2012-11-10 05:03:59 +04:00
marina.kolpakova	08b4e780de	add shrinking kernel	2012-11-10 05:03:29 +04:00
marina.kolpakova	0691dc554f	fix compilation	2012-11-10 05:02:52 +04:00
marina.kolpakova	5d15e4ea58	CUDA kernels interface	2012-11-10 05:02:37 +04:00
marina.kolpakova	14a0dd8c98	fix typo in cascade loading	2012-11-10 05:02:01 +04:00
marina.kolpakova	a3af5ede80	CUDA callers	2012-11-10 05:01:44 +04:00
marina.kolpakova	4aac1444ad	integral channel storage are cached as a cascade's field	2012-11-10 05:01:25 +04:00
marina.kolpakova	1ab7af6995	GPU soft cascade: buffers preallocation	2012-11-10 05:01:09 +04:00
marina.kolpakova	2b7ce8b160	precompute feature response for scaling factor	2012-11-10 05:00:53 +04:00
marina.kolpakova	267d140bfe	soft cascade: gpu representation	2012-11-10 05:00:33 +04:00
marina.kolpakova	dd9c53497b	GPU interface for soft cascade	2012-11-10 05:00:16 +04:00
marina.kolpakova	d65b3e0617	fix warning in CUDA samples	2012-11-10 04:59:55 +04:00