72 lines
2.6 KiB
Python
72 lines
2.6 KiB
Python
import os
|
|
|
|
import numpy as np
|
|
from mmpose.apis import MMPoseInferencer
|
|
|
|
# ==================================================================================================
|
|
|
|
filepath = os.path.dirname(os.path.realpath(__file__)) + "/"
|
|
|
|
# ==================================================================================================
|
|
|
|
|
|
def load_model():
|
|
print("Loading mmpose model ...")
|
|
|
|
model = MMPoseInferencer(
|
|
pose2d="/mmpose/projects/rtmpose/rtmpose/body_2d_keypoint/rtmpose-m_8xb256-420e_coco-384x288.py",
|
|
pose2d_weights="https://download.openmmlab.com/mmpose/v1/projects/rtmposev1/rtmpose-m_simcc-body7_pt-body7_420e-384x288-65e718c4_20230504.pth",
|
|
det_model="/mmpose/projects/rtmpose/rtmdet/person/rtmdet_nano_320-8xb32_coco-person.py",
|
|
det_weights="https://download.openmmlab.com/mmpose/v1/projects/rtmpose/rtmdet_nano_8xb32-100e_coco-obj365-person-05d8511e.pth",
|
|
det_cat_ids=[0],
|
|
)
|
|
|
|
print("Loaded mmpose model")
|
|
return model
|
|
|
|
|
|
def load_wb_model():
|
|
print("Loading mmpose whole body model ...")
|
|
|
|
model = MMPoseInferencer(
|
|
pose2d="/mmpose/projects/rtmpose/rtmpose/wholebody_2d_keypoint/rtmpose-l_8xb32-270e_coco-wholebody-384x288.py",
|
|
pose2d_weights="https://download.openmmlab.com/mmpose/v1/projects/rtmposev1/rtmpose-l_simcc-coco-wholebody_pt-aic-coco_270e-384x288-eaeb96c8_20230125.pth",
|
|
det_model="/mmpose/projects/rtmpose/rtmdet/person/rtmdet_nano_320-8xb32_coco-person.py",
|
|
det_weights="https://download.openmmlab.com/mmpose/v1/projects/rtmpose/rtmdet_nano_8xb32-100e_coco-obj365-person-05d8511e.pth",
|
|
det_cat_ids=[0],
|
|
)
|
|
|
|
print("Loaded mmpose model")
|
|
return model
|
|
|
|
|
|
# ==================================================================================================
|
|
|
|
|
|
def get_2d_pose(model, imgs, num_joints=17):
|
|
"""See: https://mmpose.readthedocs.io/en/latest/user_guides/inference.html#basic-usage"""
|
|
|
|
result_generator = model(imgs, show=False)
|
|
new_poses = []
|
|
for _ in range(len(imgs)):
|
|
result = next(result_generator)
|
|
|
|
poses = []
|
|
for i in range(len(result["predictions"][0])):
|
|
kpts = result["predictions"][0][i]["keypoints"]
|
|
scores = result["predictions"][0][i]["keypoint_scores"]
|
|
|
|
kpts = np.array(kpts)
|
|
scores = np.array(scores).reshape(-1, 1)
|
|
scores = np.clip(scores, 0, 1)
|
|
pose = np.concatenate((kpts, scores), axis=-1)
|
|
poses.append(pose)
|
|
|
|
if len(poses) == 0:
|
|
poses.append(np.zeros([num_joints, 3]))
|
|
|
|
poses = np.array(poses)
|
|
new_poses.append(poses)
|
|
|
|
return new_poses
|