Comparison between Large Displacement Optical Flow Algorithms - Revision history

imported>Projects221: /* Deep Matching (4). quasi-dense correspondes */

2014-03-23T00:01:29Z

Deep Matching (4). quasi-dense correspondes

← Older revision		Revision as of 00:01, 23 March 2014
Line 38:		Line 38:
	[[File:aggregationdetail.png \|center\|frame \| Figure 4. More detailed procedure on aggregation step (This figure is excerpted from [5]) ]]		[[File:aggregationdetail.png \|center\|frame \| Figure 4. More detailed procedure on aggregation step (This figure is excerpted from [5]) ]]

	=== Deep Matching (4). quasi-dense ~~correspondes~~ ===		=== Deep Matching (4). quasi-dense correspondences ===
	In the pyramid of multi-scale response maps, local maximum is obtained from all the matched patch, even in the areas with less textures or patterns. This is the aspect that makes Deep Matching and Deep Flow stronger than Brox and Malik's method which uses sparse descriptor matching and tends to miss the points in the textureless areas.		In the pyramid of multi-scale response maps, local maximum is obtained from all the matched patch, even in the areas with less textures or patterns[5]. This is the aspect that makes Deep Matching and Deep Flow stronger than Brox and Malik's method[3] which uses sparse descriptor matching and tends to miss the points in the textureless areas.

	[[File:multiscalepyramid.png \|center\|frame \| Figure 5. Multi-scale response map pyramid and quasi-dense correspondences (This figure is also excerpted from [5]) ]]		[[File:multiscalepyramid.png \|center\|frame \| Figure 5. Multi-scale response map pyramid and quasi-dense correspondences (This figure is also excerpted from [5]) ]]

imported>Projects221: /* Deep Matching (3). aggregating into larger patches */

2014-03-22T23:58:44Z

Deep Matching (3). aggregating into larger patches

← Older revision		Revision as of 23:58, 22 March 2014
Line 32:		Line 32:

	=== Deep Matching (3). aggregating into larger patches ===		=== Deep Matching (3). aggregating into larger patches ===
	After response maps are generated from convolution, do max-pooling and sub-sample the response map to reduce the size into half and aggregate the patches so create 8 by 8 patches from 4 by 4 response maps. Repeat this process until 16 by 16 patches and 32 by 32 patches are obtained. This multi-layer structure is similar to deep convolutional nets[6] and that is the reason the name Deep Matching and Deep Flow comes from.		After response maps are generated from convolution, do max-pooling and sub-sample the response map to reduce the size into half and aggregate the patches so create 8 by 8 patches from 4 by 4 response maps[5]. Repeat this process until 16 by 16 patches and 32 by 32 patches are obtained[5]. This multi-layer structure is similar to deep convolutional nets[6] and that is the reason the name Deep Matching and Deep Flow comes from[5].

	[[File:aggregation.png\|center\|frame \| Figure 3. Procedure to make multi-scale response maps (This figure is excerpted from [5])]]		[[File:aggregation.png\|center\|frame \| Figure 3. Procedure to make multi-scale response maps (This figure is excerpted from [5])]]

imported>Projects221: /* Deep Matching (2). convolution */

2014-03-22T23:56:40Z

Deep Matching (2). convolution

← Older revision		Revision as of 23:56, 22 March 2014
Line 27:		Line 27:

	=== Deep Matching (2). convolution ===		=== Deep Matching (2). convolution ===
	The reference(first) image is divided into non-overlapping patches of 4 by 4 pixels and convoluted with the target(second) image. And the result of convolution is the response map.		The reference(first) image is divided into non-overlapping patches of 4 by 4 pixels and convoluted with the target(second) image[5]. And the result of convolution is the response map[5].

	[[File:convolution.png\|center\|frame \| Figure 2. convolution of reference image patches and target image (This figure is excerpted from [5]) ]]		[[File:convolution.png\|center\|frame \| Figure 2. convolution of reference image patches and target image (This figure is excerpted from [5]) ]]

imported>Projects221: /* Deep Matching (1). independently movable subpatches */

2014-03-22T23:55:54Z

Deep Matching (1). independently movable subpatches

← Older revision		Revision as of 23:55, 22 March 2014
Line 22:		Line 22:

	=== Deep Matching (1). independently movable subpatches ===		=== Deep Matching (1). independently movable subpatches ===
	As we can see from the figure below, in SIFT descriptor matching, for example in the figure below using the conventional HoG template matching, the second configuration is the best matching result. However, rather than using rigid configuration of each subpatches, Deep Matching allows each subpatch to to find the best fit to the target image from the reference image.		As we can see from the figure below, in SIFT descriptor matching, for example in the figure below using the conventional HoG template matching, the second configuration is the best matching result. However, rather than using rigid configuration of each subpatches, Deep Matching allows each subpatch to to find the best fit to the target image from the reference image[5].

	[[File:subpatches.png \|center\|frame \| Figure 1. Movable cells in SIFT descriptor. Left: The reference image, Center: optimal matching in target image with fixed configuration of SIFT, Right: optimal matching on the target image with movable subpatches (This figure is excerpted from [5])]]		[[File:subpatches.png \|center\|frame \| Figure 1. Movable cells in SIFT descriptor. Left: The reference image, Center: optimal matching in target image with fixed configuration of SIFT, Right: optimal matching on the target image with movable subpatches (This figure is excerpted from [5])]]

imported>Projects221: /* Methods - Deep Flow and Deep Matching */

2014-03-22T23:54:50Z

Methods - Deep Flow and Deep Matching

← Older revision		Revision as of 23:54, 22 March 2014
Line 18:		Line 18:
	=== Deep Flow ===		=== Deep Flow ===
	<br>		<br>
	In [5], Deep Flow has been suggested that uses Deep Matching in descriptor matching. In spirit, Deep Flow is similar to Brox and Malik's work[3] in several aspects that Deep Flow also uses variational flow in their energy minimization and incorporated descriptor matching to find the corresponding regions, but it is different in that Brox and Malik[3] does sparse descriptor matching and Deep Flow is doing dense correspondences matching, what they call Deep Matching. In following sections we explain more about Deep Flow properties, structure and formula.		In [5], Deep Flow has been suggested that uses Deep Matching in descriptor matching. In spirit, Deep Flow is similar to Brox and Malik's work[3] in several aspects that Deep Flow also uses variational flow in their energy minimization and incorporated descriptor matching to find the corresponding regions, but it is different in that Brox and Malik[3] does sparse descriptor matching and Deep Flow is doing dense correspondences matching[5], what they call Deep Matching. In following sections we explain more about Deep Flow properties, structure and formula.
	<br>		<br>

imported>Projects221: /* Results */

2014-03-22T23:49:27Z

Results

← Older revision		Revision as of 23:49, 22 March 2014
Line 50:		Line 50:
	We also ran the code from [8] and compared the optical flow computation result. And the optical flow is visualized like following. Xu et al. proposed their motion detail preserving algorithm[8] to keep both small displacement optical flow and large displacement optical flow especially when there exists camera motion because of the large displacement of the background the fine structure or small displacement of the foreground is ignored. In this example, Deep Flow also gave a pretty good result to find out the motion of the arms and head but it seems like the result by [8] gives more clear holistic shape of the person on the right and give more detailed information about the motion in the arm of the person on the right.		We also ran the code from [8] and compared the optical flow computation result. And the optical flow is visualized like following. Xu et al. proposed their motion detail preserving algorithm[8] to keep both small displacement optical flow and large displacement optical flow especially when there exists camera motion because of the large displacement of the background the fine structure or small displacement of the foreground is ignored. In this example, Deep Flow also gave a pretty good result to find out the motion of the arms and head but it seems like the result by [8] gives more clear holistic shape of the person on the right and give more detailed information about the motion in the arm of the person on the right.
	<br>		<br>
	[[File:mdpofresultsmall.png\|center\|frame \| Figure 7. optical flow visualization computed with motion detail preserving method[8] ]]		[[File:mdpofresultsmall.png\|center\|frame \| Figure 7. optical flow visualization computed with motion detail preserving method[8] on the same two consecutive frames as in Figure 6. ]]

	= Conclusions and Future work =		= Conclusions and Future work =

imported>Projects221: /* Results */

2014-03-22T23:47:23Z