- Download CUDA 7.5 from https://developer.nvidia.com/cuda-downloads and install .deb package
- sudo apt-get update
- sudo apt-get install cuda
- sudo apt-get install libdc1394-22-dev
- sudo apt-get install libva-dev
With OwlCompress, users can compress a video to achieve a smaller bitrate, while maintaining high quality with regard to its Regions-of-Interest (ROI).Firstly, OwlCompress loads a video, and users mark a few frames as Key Frames. With Graph-Cut algorithms, users can mark out the ROIs in the Key Frames in an interactive and semi-automatic manner. With the ROIs of the Key Frames provided, OwlCompress automatically calculates all the ROIs of the remaining frames (non-Key frames). The information about the distribution of all the ROIs will be utilized for video compression transcoding. non-ROI area will be further compressed while ROI area will be enhanced comparing to non-ROI area.
- User Interface
First, Start the software by running “start-linux.sh”.
In OwlCompress, the user interface can be categorized as 4 areas: Main UI on the left side, Configuration Menu on the right side, System Info Bar on the very bottom-left and Zoom Bar on the very bottom-right.
The Main UI can be further split into 3 parts: Toolbox on the left side, providing buttons for user interaction; Display on the right side for viewing frames and providing intuitive interaction; Timeline at the bottom providing temporal control.
- Preview Mode
After starting the OwlCompress client, users can click on the “Open” button to load a video. So far, OwlCompress can process all types of videos supported by OpenCV 3.0 video decoder. After successfully loading the video, the Main UI enters Preview Mode, and the Configuration Menu is switched to Preview Bar.In Preview Mode, the Toolbox has the following buttons:- Play: play the video.
– Frame: jump to the previous/next frame.
– Key: jump to the previous/next Key frame.
– Make/Edit Key: mark the current frame as Key Frame and start editing. If it’s already a Key Frame, switch to Editing Mode.
– Calc: start calculating the ROIs of non-Key frames.
– Transcode: compress the video using the current ROI information.
– Export: export the ROI information to .dat binary files.
– Close: close the currently loaded video. All progress will be reset.The Display shows the image of the current frame. Zoom Bar on the very bottom-right can be used to zoom in/out. In the Timeline, each vertical segment (Frame Segment) represents a frame, arranged in temporal order:
– light dash line: non-Key frame, ROI undefined;
– light solid line: non-Key frame, ROI defined;
– heavy dash line: Key frame, ROI undefined;
– heavy solid line: Key frame, ROI defined;Timestamps/frame index are shown above/below the Frame Segments, respectively. In Preview Mode, users can jump to an arbitrary frame by clicking on the corresponding Frame Segment in the Timeline.
- Editing Mode
When the “Make/Edit Key” button in Preview Mode is clicked, OwlCompress enters Editing Mode. In Editing Mode, jumping to another frame is not allowed, users can only edit the ROI of the current Key Frame.
In Editing Mode, the Toolbox has the following buttons:- Foreground: mark foreground area in the current frame;
– Background: mark background area in the current frame;
– Domain: use rectangles to define possible foreground area. All area outside of domain will be excluded from ROI.
– Erase: erase any markers.
– Generate Mask: manually run the ROI calculation.
– Hide Indicators: hide all markers, only shows the ROI.
– Reset Calculation: reset all progress of the current frame.
– Return to Preview
– Delete Key: delete this Key Frame and return to Preview Mode.In Editing Mode, users first use “Domain” tool to specify the possible area of ROIs, and then use “Foreground” or “Background” to mark the ROI/non-ROI area. After each operation, OwlCompress will automatically calculate the ROI area.
- Calculation Mode
If all Key Frames have their ROIs defined, clicking “Calc” in Preview Mode Toolbox sets OwlCompress to Calculation Mode.
In Calculation Mode, the Toolbox has the following buttons:- Start: start the calculation of the ROIs of non-Key frames.
– Stop: pause the calculation.
– Reset: reset the calculation of the current frame.
– Reset All: reset the calculations of all non-Key frames.
- Configuration Menu
The Configuration Menu has 4 tabs, “I/O”, “Preview”, “Edit” and “Advanced”. They provide user configurations to the following parameters:- I/O: input/output location, quality parameter (qp) for enhancing ROI area, qp for fading non-ROI area, encoding method.
– Preview: video play speed, cache size for previewing, neighbor size for consecutive caching.
– Edit: brush size, erase size, mask color, mask transparency.
– Advanced: advanced parameters, including downscale values (larger for faster but rough calculations) and tracking steps (granularity for ROI tracking, smaller for more frequent tracking).