blocks|key|2523688|text|使用卡尔曼滤波器的目的是为了目标跟踪(不确定是否平滑)。如果C%2B%2B实现还可以，那么您可能想要使用这个流行的github存储库https://github.com/AlexeyAB/darknet|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|2523689|如果你阅读了文档，它有C%2B%2B+API，你可以使用darknet作为库(这样你就可以使用你的yolo模型)并将其加载到你的C%2B%2B程序中。请看一下在https://github.com/AlexeyAB/darknet/blob/master/src/yolo_console_dll.cpp中使用darknet库C%2B%2B程序示例。|2523690|在C%2B%2B代码中，作者给出了3种进行目标跟踪选项，其中1种是使用卡尔曼滤波：|2523691|2523692|Track光流算法，但它只适用于实时检测，不适用于视频。您可以通过取消注释此行//#define+TRACK_OPTFLOW来使用算法。看一下508+~+522行|ordered-list-item|style|BOLD|CODE|2523693|2523694|#ifdef+TRACK_OPTFLOW
++++++++++++++++++++++++if+(detection_data.new_detection)+{
++++++++++++++++++++++++++++tracker_flow.update_tracking_flow(detection_data.cap_frame,+detection_data.result_vec);
++++++++++++++++++++++++++++while+(track_optflow_queue.size()+>+0)+{
++++++++++++++++++++++++++++++++draw_frame+=+track_optflow_queue.back();
++++++++++++++++++++++++++++++++result_vec+=+tracker_flow.tracking_flow(track_optflow_queue.front(),+false);
++++++++++++++++++++++++++++++++track_optflow_queue.pop();
++++++++++++++++++++++++++++}
++++++++++++++++++++++++}
++++++++++++++++++++++++else+{
++++++++++++++++++++++++++++track_optflow_queue.push(cap_frame);
++++++++++++++++++++++++++++result_vec+=+tracker_flow.tracking_flow(cap_frame,+false);
++++++++++++++++++++++++}
++++++++++++++++++++++++detection_data.new_detection+=+true;++++//+to+correct+kalman+filter
#endif+//TRACK_OPTFLOW|code-block|syntax|javascript|2523695|2523696|Kalman+Filter，不是真正推荐的，因为它不是真的准确，但可能适用于闭路电视或固定摄像头。要使用卡尔曼滤波器，请将此值更改为true+bool+const+use_kalman_filter+=+false;。看一下524+~532行|2523697|2523698|//+track+ID+by+using+kalman+filter
++++++++++++++++++++++++if+(use_kalman_filter)+{
++++++++++++++++++++++++++++if+(detection_data.new_detection)+{
++++++++++++++++++++++++++++++++result_vec+=+track_kalman.correct(result_vec);
++++++++++++++++++++++++++++}
++++++++++++++++++++++++++++else+{
++++++++++++++++++++++++++++++++result_vec+=+track_kalman.predict();
++++++++++++++++++++++++++++}
++++++++++++++++++++++++}|2523699|2523700|//+track+ID+by+using+custom+function
++++++++++++++++++++++++else+{
++++++++++++++++++++++++++++int+frame_story+=+std::max(5,+current_fps_cap.load());
++++++++++++++++++++++++++++result_vec+=+detector.tracking_id(result_vec,+true,+frame_story,+40);
++++++++++++++++++++++++}|2523701|entityMap|0|LINK|mutability|MUTABLE|url|https://github.com/AlexeyAB/darknet|1|https://github.com/AlexeyAB/darknet/blob/master/src/yolo_console_dll.cpp^0|1Q|Z|0|0|20|20|1|0|0|0|0|7|13|N|0|0|0|0|0|D|1Z|11|0|0|0|0|0^^$0|@$1|2|3|4|5|6|7|1D|8|@]|9|@$A|1E|B|1F|1|1G]]|C|$]]|$1|D|3|E|5|6|7|1H|8|@]|9|@$A|1I|B|1J|1|1K]]|C|$]]|$1|F|3|G|5|6|7|1L|8|@]|9|@]|C|$]]|$1|H|3|-4|5|6|7|1M|8|@]|9|@]|C|$]]|$1|I|3|J|5|K|7|1N|8|@$A|1O|B|1P|L|M]|$A|1Q|B|1R|L|N]]|9|@]|C|$]]|$1|O|3|-4|5|6|7|1S|8|@]|9|@]|C|$]]|$1|P|3|Q|5|R|7|1T|8|@]|9|@]|C|$S|T]]|$1|U|3|-4|5|6|7|1U|8|@]|9|@]|C|$]]|$1|V|3|W|5|K|7|1V|8|@$A|1W|B|1X|L|M]|$A|1Y|B|1Z|L|N]]|9|@]|C|$]]|$1|X|3|-4|5|6|7|20|8|@]|9|@]|C|$]]|$1|Y|3|Z|5|R|7|21|8|@]|9|@]|C|$S|T]]|$1|10|3|-4|5|6|7|22|8|@]|9|@]|C|$]]|$1|11|3|12|5|R|7|23|8|@]|9|@]|C|$S|T]]|$1|13|3|-4|5|6|7|24|8|@]|9|@]|C|$]]]|14|$15|$5|16|17|18|C|$19|1A]]|1B|$5|16|17|18|C|$19|1C]]]]

The purpose of using Kalman filter is for object tracking (not sure for smoothing). If C++ implementation is okay, You probably want to use this popular github repository <a href="https://github.com/AlexeyAB/darknet" rel="nofollow noreferrer">https://github.com/AlexeyAB/darknet</a>

If you read the documentation, it has C++ API which you can use darknet as library (so you can use your yolo model) and load it to your C++ program. Take a look at C++ program example that use darknet library here <a href="https://github.com/AlexeyAB/darknet/blob/master/src/yolo_console_dll.cpp" rel="nofollow noreferrer">https://github.com/AlexeyAB/darknet/blob/master/src/yolo_console_dll.cpp</a> .

In that C++ code, the author gives 3 option to do object tracking, 1 of them is using kalman filter : 

<ol>
<li>Track Optical Flow algorithm, but it only works for live detection, not for video. You can use the algorithm by uncomment this line <code>//#define TRACK_OPTFLOW</code> . Take a look at line 508 ~ 522 </li>
</ol>

<pre><code>#ifdef TRACK_OPTFLOW
 if (detection_data.new_detection) {
 tracker_flow.update_tracking_flow(detection_data.cap_frame, detection_data.result_vec);
 while (track_optflow_queue.size() &gt; 0) {
 draw_frame = track_optflow_queue.back();
 result_vec = tracker_flow.tracking_flow(track_optflow_queue.front(), false);
 track_optflow_queue.pop();
 }
 }
 else {
 track_optflow_queue.push(cap_frame);
 result_vec = tracker_flow.tracking_flow(cap_frame, false);
 }
 detection_data.new_detection = true; // to correct kalman filter
#endif //TRACK_OPTFLOW
</code></pre>

<ol start="2">
<li>Kalman Filter, not really reccommended because it's not really accurate, but might work for CCTV or stationary camera. To use kalman filter change this value to true <code>bool const use_kalman_filter = false;</code>. Take a look at line 524 ~ 532</li>
</ol>

<pre><code>// track ID by using kalman filter
 if (use_kalman_filter) {
 if (detection_data.new_detection) {
 result_vec = track_kalman.correct(result_vec);
 }
 else {
 result_vec = track_kalman.predict();
 }
 }
</code></pre>

<ol start="3">
<li>Custom Object Tracker, I used this custom function and it performs better than kalman filter in my case, it will give you track ID for each object. </li>
</ol>

<pre><code>// track ID by using custom function
 else {
 int frame_story = std::max(5, current_fps_cap.load());
 result_vec = detector.tracking_id(result_vec, true, frame_story, 40);
 }
</code></pre>

I would like to make people detection smooth by adding Kalman filter to YOLO.

In another post I noticed* (see code below in image.c file) how to get the coordinates of the bounding box in YOLO.

<pre><code>if(bot &gt; im.h-1) bot = im.h-1;

// Print bounding box values 
printf("Bounding Box: Left=%d, Top=%d, Right=%d, Bottom=%d\n", left, top, right, bot); 
draw_box_width(im, left, top, right, bot, width, red, green, blue);
</code></pre>

How to add Kalman filter to YOLO v2

我想通过在YOLO中添加卡尔曼滤波器来使人的检测变得平滑。在另一篇文章中，我注意到*(参见下面image.c文件中的代码)如何在YOLO中获取边界框的坐标。if(bot > im.h-1) bot = im.h-1;// Print bounding box values printf("Bounding Box: L...

问如何在YOLO v2中添加卡尔曼滤波器
EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问如何在YOLO v2中添加卡尔曼滤波器EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问如何在YOLO v2中添加卡尔曼滤波器
EN