The site is read-only. Please transition to use Robotics Stack Exchange
| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
 | 1 | initial version |
You do a lot of work in your callback, and your subscribe queue is limited to one image. It is not surprising that frames get dropped while your callback is busy with your classifier.
As I see it, you have three options:
Ignore the frame drops, and process just those frames you can actually handle.
Increase the queue size. This will only work if you can keep up with the frame rate most of the time. Otherwise, your queue will just fill up and your node will lag behind accordingly.
Use a faster classifier.
| | 2 | No.2 Revision |
You do a lot of work in your callback, and your subscribe queue is limited to one image. It is not surprising that frames get dropped while your callback is busy with your classifier.
As I see it, you have three options:
Ignore the frame drops, and process just those frames you can actually handle.
Increase the queue size. This will only work if you can keep up with the frame rate most of the time. Otherwise, your queue will just fill up and your node will lag behind accordingly.
Use a faster Improve upon the speed of your classifier.
| | 3 | No.3 Revision |
You do a lot of work in your callback, and your subscribe queue is limited to one image. It is not surprising that frames get dropped while your callback is busy with your classifier.
As I see it, you have three options:
Ignore the frame drops, and process just those frames you can actually handle.
Increase the queue size. This will only work if you can keep up with the frame rate most of the time. Otherwise, your queue will just fill up and your node will lag behind accordingly.
Improve upon the speed of your classifier.classifier. Use a faster algorithm and/or reduce the image size.
| | 4 | No.4 Revision |
You do a lot of work in your callback, and your subscribe queue is limited to one image. It is not surprising that frames get dropped while your callback is busy with your classifier.
As I see it, you have three options:
Ignore the frame drops, and process just those frames you can actually handle.
Increase the queue size. This will only work if you can keep up with the frame rate most of the time. Otherwise, your queue will just fill up and your node will lag behind accordingly.
Do less work in your callback. Improve the speed of your classifier. Use a faster algorithm and/or reduce the image size.classifier.
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
