From 9460caec9d3df234d35b986d69958e68167ecd8c Mon Sep 17 00:00:00 2001
From: Wei-Chen-hub <1259566226@qq.com>
Date: Tue, 15 Aug 2023 14:06:42 +0800
Subject: [PATCH] update 230815

---
 tools/postprocess/humandata_sample.py | 103 ++++++++++++++++++++++++++
 1 file changed, 103 insertions(+)

diff --git a/tools/postprocess/humandata_sample.py b/tools/postprocess/humandata_sample.py
index e69de29b..f2aa6504 100644
--- a/tools/postprocess/humandata_sample.py
+++ b/tools/postprocess/humandata_sample.py
@@ -0,0 +1,103 @@
+import argparse
+import glob
+import os
+import pdb
+
+from tqdm import tqdm
+
+from mmhuman3d.data.data_structures.human_data import HumanData
+
+
+def sample_humandata_from_key_param(src, dst, key, param) -> None:
+    """This function is used to sample selected HumanData from all HumanData,
+    based on given keys and parameters.
+
+    Inputs:
+        - src: str, the path of the source HumanData (glob.glob syntax)
+        - dst: str, output path
+        - key: str, the key of the HumanData to be sampled according to
+        - param: str, the parameter of sample criteria
+
+    Returns:
+        - None
+    """
+
+    # prepare humandata paths
+    src_ps = glob.glob(src)
+
+    # init humandata
+    # human_data_s = HumanData()
+
+    # sample
+    slices = []
+    for src_idx, srcp in enumerate(src_ps):
+        human_data = HumanData.fromfile(srcp)
+
+        for idx, k in enumerate(
+                tqdm(human_data[key]),
+                desc=f'Sampling from npz files {src_idx + 1} / {len(src_ps)}'):
+            if k in param:
+                selected_slice = human_data.get_slice(idx, idx + 1)
+                slices.append(selected_slice)
+
+    # concat slices to humandata
+    slice_sample = selected_slice[0]
+    # keys = list(slice_sample.keys())
+
+    human_data_s = slice_sample
+    for s in tqdm(slices[1:], desc='Concatenating slices'):
+        human_data_s = human_data_s.concatenate(human_data_s, s)
+
+    # save humandata
+    human_data_s.dump(dst)
+
+
+def main(args):
+    """Sample a humandata from humandata based on humandata Currently support
+    sampling based on image_path."""
+    base_dir = args.humandata_dir
+    output_dir = args.output_dir
+    fn = args.humandata_sample_base
+
+    # load base humandata
+    base_ps = glob.glob(os.path.join(base_dir, fn))
+    criteria = args.key
+    c_param = []
+    for bp in base_ps:
+        human_data = HumanData.fromfile(bp)
+
+        c_param += human_data[criteria]
+
+    # prepare to sample
+    sample_humandata_from_key_param(
+        src=os.path.join(base_dir, args.humandata_src),
+        dst=os.path.join(output_dir, fn).replace('.npz',
+                                                 f'_{criteria}_sampled.npz'),
+        key=criteria,
+        param=c_param)
+
+    pdb.set_trace()
+
+
+if __name__ == '__main__':
+
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        '--humandata_dir',
+        type=str,
+        default='/mnt/c/Users/12595/Desktop/synbody_liushuai')
+    parser.add_argument(
+        '--output_dir',
+        type=str,
+        default='/mnt/c/Users/12595/Desktop/synbody_liushuai')
+
+    parser.add_argument(
+        '--humandata_src', type=str, default='synbody_v1*38400*.npz')
+    parser.add_argument(
+        '--humandata_sample_base', type=str, default='synbody_v1_1_10w.npz')
+
+    parser.add_argument('--key', type=str, default='image_path')
+    args = parser.parse_args()
+
+    main(args)