Effective Optical Flow for small displacements
.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty{ height:90px;width:728px;box-sizing:border-box;
}
I am working on a project where I must try to recognize tad movements around the nose, mouth, eyes. Movements which take ms. I am working with OpenCV 3.4 and Python 3 respectively. Currently I am taking the Dense Optical Flow of 300x300 frame cropped from the original 1080p one.
The problem is that the performance is seriously hurt as I am running the algorithm at around 15fps. I started thinking about switching to the sparse lucas-kanade approach and making just clouds of points where needed.
I need an educated advice about how to tackle the problem.
- Is it better to switch to LK Optical Flow or rather stick to the Dense one.
- Is it worth downscaling (with pyramid) for the Dense OF that 300x300
image or that will loose me the small movements? Shall I distribute the calculations between two cores? - How can I evaluate optical flow's output?
Essentially, which is the dense or the sparse approach better for this scenario and what do you advice me to do to strike the balance between accuracy and performance. Even window-size or number iterations in Farneback will be helpful if you can tell whether or not tweaking them is a good decision.
python opencv image-processing
add a comment |
I am working on a project where I must try to recognize tad movements around the nose, mouth, eyes. Movements which take ms. I am working with OpenCV 3.4 and Python 3 respectively. Currently I am taking the Dense Optical Flow of 300x300 frame cropped from the original 1080p one.
The problem is that the performance is seriously hurt as I am running the algorithm at around 15fps. I started thinking about switching to the sparse lucas-kanade approach and making just clouds of points where needed.
I need an educated advice about how to tackle the problem.
- Is it better to switch to LK Optical Flow or rather stick to the Dense one.
- Is it worth downscaling (with pyramid) for the Dense OF that 300x300
image or that will loose me the small movements? Shall I distribute the calculations between two cores? - How can I evaluate optical flow's output?
Essentially, which is the dense or the sparse approach better for this scenario and what do you advice me to do to strike the balance between accuracy and performance. Even window-size or number iterations in Farneback will be helpful if you can tell whether or not tweaking them is a good decision.
python opencv image-processing
2
You need to share us some images for us to provide you with proper answer. All of the decisions are based on how sensitive the motions you are planning to detect. If the motion is large enough to be detected at smaller scale, then going to the smaller scale is the way to go. If there is no way of reducing size, then you will need to reduce the process area by running the algorithm on a specific ROI instead of the entire image.
– yapws87
Nov 24 '18 at 8:23
add a comment |
I am working on a project where I must try to recognize tad movements around the nose, mouth, eyes. Movements which take ms. I am working with OpenCV 3.4 and Python 3 respectively. Currently I am taking the Dense Optical Flow of 300x300 frame cropped from the original 1080p one.
The problem is that the performance is seriously hurt as I am running the algorithm at around 15fps. I started thinking about switching to the sparse lucas-kanade approach and making just clouds of points where needed.
I need an educated advice about how to tackle the problem.
- Is it better to switch to LK Optical Flow or rather stick to the Dense one.
- Is it worth downscaling (with pyramid) for the Dense OF that 300x300
image or that will loose me the small movements? Shall I distribute the calculations between two cores? - How can I evaluate optical flow's output?
Essentially, which is the dense or the sparse approach better for this scenario and what do you advice me to do to strike the balance between accuracy and performance. Even window-size or number iterations in Farneback will be helpful if you can tell whether or not tweaking them is a good decision.
python opencv image-processing
I am working on a project where I must try to recognize tad movements around the nose, mouth, eyes. Movements which take ms. I am working with OpenCV 3.4 and Python 3 respectively. Currently I am taking the Dense Optical Flow of 300x300 frame cropped from the original 1080p one.
The problem is that the performance is seriously hurt as I am running the algorithm at around 15fps. I started thinking about switching to the sparse lucas-kanade approach and making just clouds of points where needed.
I need an educated advice about how to tackle the problem.
- Is it better to switch to LK Optical Flow or rather stick to the Dense one.
- Is it worth downscaling (with pyramid) for the Dense OF that 300x300
image or that will loose me the small movements? Shall I distribute the calculations between two cores? - How can I evaluate optical flow's output?
Essentially, which is the dense or the sparse approach better for this scenario and what do you advice me to do to strike the balance between accuracy and performance. Even window-size or number iterations in Farneback will be helpful if you can tell whether or not tweaking them is a good decision.
python opencv image-processing
python opencv image-processing
edited Nov 23 '18 at 23:32
KDX2
asked Nov 23 '18 at 23:12
KDX2KDX2
3342318
3342318
2
You need to share us some images for us to provide you with proper answer. All of the decisions are based on how sensitive the motions you are planning to detect. If the motion is large enough to be detected at smaller scale, then going to the smaller scale is the way to go. If there is no way of reducing size, then you will need to reduce the process area by running the algorithm on a specific ROI instead of the entire image.
– yapws87
Nov 24 '18 at 8:23
add a comment |
2
You need to share us some images for us to provide you with proper answer. All of the decisions are based on how sensitive the motions you are planning to detect. If the motion is large enough to be detected at smaller scale, then going to the smaller scale is the way to go. If there is no way of reducing size, then you will need to reduce the process area by running the algorithm on a specific ROI instead of the entire image.
– yapws87
Nov 24 '18 at 8:23
2
2
You need to share us some images for us to provide you with proper answer. All of the decisions are based on how sensitive the motions you are planning to detect. If the motion is large enough to be detected at smaller scale, then going to the smaller scale is the way to go. If there is no way of reducing size, then you will need to reduce the process area by running the algorithm on a specific ROI instead of the entire image.
– yapws87
Nov 24 '18 at 8:23
You need to share us some images for us to provide you with proper answer. All of the decisions are based on how sensitive the motions you are planning to detect. If the motion is large enough to be detected at smaller scale, then going to the smaller scale is the way to go. If there is no way of reducing size, then you will need to reduce the process area by running the algorithm on a specific ROI instead of the entire image.
– yapws87
Nov 24 '18 at 8:23
add a comment |
0
active
oldest
votes
Your Answer
StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53453764%2feffective-optical-flow-for-small-displacements%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
0
active
oldest
votes
0
active
oldest
votes
active
oldest
votes
active
oldest
votes
Thanks for contributing an answer to Stack Overflow!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53453764%2feffective-optical-flow-for-small-displacements%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
2
You need to share us some images for us to provide you with proper answer. All of the decisions are based on how sensitive the motions you are planning to detect. If the motion is large enough to be detected at smaller scale, then going to the smaller scale is the way to go. If there is no way of reducing size, then you will need to reduce the process area by running the algorithm on a specific ROI instead of the entire image.
– yapws87
Nov 24 '18 at 8:23