PaddleDetection-release-2.6/deploy/python/det_keypoint_unite_utils.py

# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import ast
import argparse


def argsparser():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument(
        "--det_model_dir",
        type=str,
        default=None,
        help=("Directory include:'model.pdiparams', 'model.pdmodel', "
              "'infer_cfg.yml', created by tools/export_model.py."),
        required=True)
    parser.add_argument(
        "--keypoint_model_dir",
        type=str,
        default=None,
        help=("Directory include:'model.pdiparams', 'model.pdmodel', "
              "'infer_cfg.yml', created by tools/export_model.py."),
        required=True)
    parser.add_argument(
        "--image_file", type=str, default=None, help="Path of image file.")
    parser.add_argument(
        "--image_dir",
        type=str,
        default=None,
        help="Dir of image file, `image_file` has a higher priority.")
    parser.add_argument(
        "--keypoint_batch_size",
        type=int,
        default=8,
        help=("batch_size for keypoint inference. In detection-keypoint unit"
              "inference, the batch size in detection is 1. Then collate det "
              "result in batch for keypoint inference."))
    parser.add_argument(
        "--video_file",
        type=str,
        default=None,
        help="Path of video file, `video_file` or `camera_id` has a highest priority."
    )
    parser.add_argument(
        "--camera_id",
        type=int,
        default=-1,
        help="device id of camera to predict.")
    parser.add_argument(
        "--det_threshold", type=float, default=0.5, help="Threshold of score.")
    parser.add_argument(
        "--keypoint_threshold",
        type=float,
        default=0.5,
        help="Threshold of score.")
    parser.add_argument(
        "--output_dir",
        type=str,
        default="output",
        help="Directory of output visualization files.")
    parser.add_argument(
        "--run_mode",
        type=str,
        default='paddle',
        help="mode of running(paddle/trt_fp32/trt_fp16/trt_int8)")
    parser.add_argument(
        "--device",
        type=str,
        default='cpu',
        help="Choose the device you want to run, it can be: CPU/GPU/XPU, default is CPU."
    )
    parser.add_argument(
        "--run_benchmark",
        type=ast.literal_eval,
        default=False,
        help="Whether to predict a image_file repeatedly for benchmark")
    parser.add_argument(
        "--enable_mkldnn",
        type=ast.literal_eval,
        default=False,
        help="Whether use mkldnn with CPU.")
    parser.add_argument(
        "--cpu_threads", type=int, default=1, help="Num of threads with CPU.")
    parser.add_argument(
        "--trt_min_shape", type=int, default=1, help="min_shape for TensorRT.")
    parser.add_argument(
        "--trt_max_shape",
        type=int,
        default=1280,
        help="max_shape for TensorRT.")
    parser.add_argument(
        "--trt_opt_shape",
        type=int,
        default=640,
        help="opt_shape for TensorRT.")
    parser.add_argument(
        "--trt_calib_mode",
        type=bool,
        default=False,
        help="If the model is produced by TRT offline quantitative "
        "calibration, trt_calib_mode need to set True.")
    parser.add_argument(
        '--use_dark',
        type=ast.literal_eval,
        default=True,
        help='whether to use darkpose to get better keypoint position predict ')
    parser.add_argument(
        '--save_res',
        type=bool,
        default=False,
        help=(
            "whether to save predict results to json file"
            "1) store_res: a list of image_data"
            "2) image_data: [imageid, rects, [keypoints, scores]]"
            "3) rects: list of rect [xmin, ymin, xmax, ymax]"
            "4) keypoints: 17(joint numbers)*[x, y, conf], total 51 data in list"
            "5) scores: mean of all joint conf"))
    parser.add_argument(
        '--smooth',
        type=ast.literal_eval,
        default=False,
        help='smoothing keypoints for each frame, new incoming keypoints will be more stable.'
    )
    parser.add_argument(
        '--filter_type',
        type=str,
        default='OneEuro',
        help='when set --smooth True, choose filter type you want to use, it can be [OneEuro] or [EMA].'
    )
    return parser