-**Operating system** (`lsb_release -a` in Ubuntu):
-**Release or Debug mode**? (by defualt: release):
- Compiler (`gcc --version` in Ubuntu or VS version in Windows): 5.4.0, ... (Ubuntu); VS2015 Enterprise Update 3, VS2017 community, ... (Windows); ...?
- Any other custom CMake configuration with respect to the default version? (by default: no):
3.**3rd-party software**:
-**Caffe version**: Default from OpenPose, custom version, ...?
-**CMake version** (`cmake --version` in Ubuntu):
-**OpenCV version**: pre-compiled `apt-get install libopencv-dev` (only Ubuntu); OpenPose default (only Windows); compiled from source? If so, 2.4.9, 2.4.12, 3.1, 3.2?; ...?
4. If **GPU mode** issue:
-**CUDA version** (`cat /usr/local/cuda/version.txt` in most cases):
-**cuDNN version**:
-**GPU model** (`nvidia-smi` in Ubuntu):
5. If **CPU-only mode** issue:
-**CPU model**:
- Total **RAM memory** available:
6. If **speed performance** issue:
- Report OpenPose timing speed based on [this link](https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/doc/installation.md#profiling-speed).
@@ -235,7 +235,7 @@ To manually select the CPU Version, open CMake GUI mentioned above, and set the
- On Ubuntu, OpenPose will link against the Intel MKL version (Math Kernel Library) of Caffe. Alternatively, the user can choose his own Caffe version, by unselecting `USE_MKL` and selecting his own Caffe path.
- On Windows, it will use the default version of Caffe or one provided by the user on the CPU.
The default CPU version takes ~0.2 seconds per image on Ubuntu (~50x slower than GPU) while the MKL version provides a roughly 2x speedup at ~0.4 seconds. As of now OpenPose does not support MKL on Windows but will at a later date. Also, MKL version does not support unfixed resolution. So a folder of images of different resolutions with openpose, requires the `--net_resolution 656x368` flag for example.
The default CPU version takes ~0.2 images per second on Ubuntu (~50x slower than GPU) while the MKL version provides a roughly 2x speedup at ~0.4 images per second. As of now OpenPose does not support MKL on Windows but will at a later date. Also, MKL version does not support unfixed resolution. So a folder of images of different resolutions requires a fixed net resolution (e.g., `--net_resolution 656x368`).
The user can configure the environmental variables `MKL_NUM_THREADS` and `OMP_NUM_THREADS`. They are set at an optimum parameter level by default (i.e., to the number of threads of the machine). However, they can be tweak by running the following commands into the terminal window, right before running any OpenPose application. Eg:
5. Function `scaleKeypoints(Array<float>& keypoints, const float scale)` also accepts 3D keypoints.
6. 3D keypoints and camera parameters in meters (instead of millimeters) in order to reduce numerical errors.
7. New `PoseExtractor` class to contain future ID and tracking algorithms as well as the current OpenPose keypoint detection algorithm.
8. Added initial alpha versions of the `tracking` and `identification` modules (for now disabled but available in the source code), including `PersonIdExtractor` and `PersonTracker`. `PersonIdExtractor` includes greedy matrix OP-LK matching.
2. Functions or parameters renamed:
1. Removed scale parameter from hand and face rectangle extractor (causing wrong results if custom `--output_resolution`).
2. Functions `scaleKeypoints`, other than `scaleKeypoints(Array<float>& keypoints, const float scale)`, renamed as `scaleKeypoints2d`.
3.`(W)PoseExtractor` renamed to `(W)PoseExtractorNet` to distinguish from new `PoseExtractor`. Analogously with `(W)FaceExtractorNet` and `(W)HandExtractorNet`.
4. Experimental module removed and internal `tracking` folder moved to main openpose folder.
3. Main bugs fixed:
1. Fixed hand and face extraction and rendering scaling issues when `--output_resolution` is not the default one.
2. Part candidates (`--part_candidates`) are saved with the same scale than the final keypoints itself.
