Skip to content

Latest commit

 

History

History
38 lines (31 loc) · 2.6 KB

File metadata and controls

38 lines (31 loc) · 2.6 KB

AlphaPose Usage & Examples

Here, we first list the flags and other parameters you can tune. Default parameters work well and you don't need to tune them unless you know what you are doing.

Flags

  • --cfg: Experiment configure file name

  • --checkpoint: Experiment checkpoint file name

  • --sp: Run the program using a single process. Windows users need to turn this flag on.

  • --detector: Detector you can use, yolo/tracker.

  • --indir: Directory of the input images. All the images in the directory will be processed.

  • --list: A text file list for the input images

  • --image: Read single image and process.

  • --video: Read video and process the video frame by frame.

  • --outdir: Output directory to store the pose estimation results.

  • --vis: If turned-on, it will render the results and visualize them.

  • --save_img: If turned-on, it will render the results and save them as images in $outdir/vis.

  • --save_video: If turned-on, it will render the results and save them as a video.

  • --vis_fast: If turned on, it will use faster rendering method. Default is false.

  • --format: The format of the saved results. By default, it will save the output in COCO-like format. Alternative options are 'cmu' and 'open', which saves the results in the format of CMU-Pose or OpenPose. For more details, see output.md

  • --detbatch: Batch size for the detection network.

  • --posebatch: Maximum batch size for the pose estimation network. If you met OOM problem, decrease this value until it fit in the memory.

  • --flip: Enable flip testing. Can increase the accuracy.

  • --min_box_area: Min box area to filter out, you can set it like 100 to filter out small people.

  • --gpus: Choose which cuda device to use by index and input comma to use multi gpus, e.g. 0,1,2,3. (input -1 for cpu only)

  • --pose_track: Enable tracking pipeline with human re-id feature, it is currently the best performance pose tracker

  • --pose_flow: This flag will be depreciated. It enables the old tracking version of PoseFlow.

All the flags available here: link

Parameters

  1. yolo detector config is here
  • CONFIDENCE: Confidence threshold for human detection. Lower the value can improve the final accuracy but decrease the speed. Default is 0.05.
  • NMS_THRES: NMS threshold for human detection. Increase the value can improve the final accuracy but decrease the speed. Default is 0.6.
  • INP_DIM: The input size of detection network. The inp_dim should be multiple of 32. Default is 608. Increase it may improve the accuracy.