opentrack.git/tracker-neuralnet, branch trackhat-camerahead tracking software for Windows and Linuxhttp://ananke.misaki.pl/cgit/opentrack.git/atom?h=trackhat-camera2022-05-21T07:41:20Ztracker/nn: Fix localizer interfering with tracking when tracking is still good2022-05-21T07:41:20ZMichael Weltermichael@welter-4d.de2022-05-21T07:26:12Zurn:sha1:bbce410b03c2bb36cf2a07b1394fa04bf4f04c70tracker/nn: Fix drawing of uninitialized face ROI buffer2022-05-21T07:41:13ZMichael Weltermichael@welter-4d.de2022-05-21T07:24:43Zurn:sha1:af50614c485fede08540e8844b726b434de76d83tracker/nn: Use postfix underscore to indicate class member variables2022-05-19T18:47:11ZMichael Weltermichael@welter-4d.de2022-05-18T20:15:02Zurn:sha1:e3de47abc3eba2d1cebc94943a203623c6545f3ftracker/nn: Improve preview2022-05-19T18:47:11ZMichael Weltermichael@welter-4d.de2022-05-18T04:36:25Zurn:sha1:e1d17f2d413de5f548931eaf9dfed2155e830096
* The preview image is now generated with the dimensions of the widget
* The pose visualization is added afterwards adjusted to the preview size
* The fps / inference time readout is moved to the settings dialog
* The actual obtained resolution from the camera is also shown
* Dialog layout is changed
Note: Switching to using underscore to mark class member vars.
It's not consistently applied yet.
tracker/nn: Add resolution setting and tooltips2022-05-19T18:47:11ZMichael Weltermichael@welter-4d.de2022-05-17T05:50:55Zurn:sha1:f788d94a9ea446c7b687b4a8f3d4cfae41a61296
Tooltips only for half of the settings or so. When hovering over the
actual input boxes.
tracker/nn: Add support for recurrent model and further tweaks2022-05-19T18:46:50ZMichael Weltermichael@welter-4d.de2022-05-12T17:54:57Zurn:sha1:4d6e80a6c585f6929f6bf4bf319932732160ce84
Regarding tweaks:
* EWA smoothing of head ROI. Smoothing strength is a UI setting.
* Adjustible zoom into the detected face. The predicted ROI is scaled by a factor the user can set. There is a sweet spot somewhere near 1.
* Adjustible number of threads
* The ROI is no longer taken as model output directly. This was actually not needed. Perhaps as auxiliary training objective for the network. But the tracker implementation now just uses a square area around the head center according to the predicted head size.
* Add comment and debug notification on face ROI model output
tracker/nn: Support new coordinate conventions for head rotation2022-05-15T14:53:44ZMichael Weltermichael@welter-4d.de2022-02-26T21:20:40Zurn:sha1:25af3db8ba560d86f78aa4bfd10039588b7bab82tracker/nn: Don't hardcode the size of the input image for the pose estimator2022-05-15T14:53:20ZMichael Weltermichael@welter-4d.de2021-07-06T18:55:36Zurn:sha1:4697a0e44536ae9dc169aab54df26e5dbebdec7ctracker/nn: catch exception in case the head pose model throws during inference2022-05-12T17:55:23ZMichael Weltermichael@welter-4d.de2021-10-26T09:31:26Zurn:sha1:f58bc9de86e79df8f7a30fe1e21821bdabfb165eAdded support for MJPEG compression for neuralnet2022-05-11T18:20:14ZPedro Silvasparta14gaming@gmail.com2022-05-11T18:20:14Zurn:sha1:c4efa268c0bd783cd8b04ed1cd6601ab5d3d398a
- Added support for MJPEG compression for neuralnet tracker