diff --git a/fit/tools/cross_detector.py b/fit/tools/cross_detector.py
new file mode 100644
index 0000000..b080144
--- /dev/null
+++ b/fit/tools/cross_detector.py
@@ -0,0 +1,54 @@
+import pickle
+import numpy as np
+import os
+import json
+import argparse
+
+def parse_args():
+    parser = argparse.ArgumentParser(description='Detect cross joints')
+    parser.add_argument('--dataset_name', dest='dataset_name',
+                        help='select dataset',
+                        default='', type=str)
+    parser.add_argument('--output_path', dest='output_path',
+                        help='path of output',
+                        default=None, type=str)
+    args = parser.parse_args()
+    return args
+
+def create_dir_not_exist(path):
+    if not os.path.exists(path):
+        os.mkdir(path)
+
+def load_Jtr(file_path):
+    with open(file_path, 'rb') as f:
+        data = pickle.load(f)
+    Jtr = np.array(data["Jtr"])
+    return Jtr
+
+
+def has_cross(joints: np.ndarray):
+    return (joints[1][0]-joints[2][0]) * (joints[10][0]-joints[11][0]) < 0 or\
+           (joints[13][0]-joints[14][0]) * (joints[22][0]-joints[23][0]) < 0
+
+
+def cross_frames(Jtr: np.ndarray):
+    ans = []
+    for frame in range(Jtr.shape[0]):
+        if has_cross(Jtr[frame]):
+            ans.append(frame)
+    return ans
+
+
+def cross_detector(dir_path):
+    ans={}
+    for root, dirs, files in os.walk(dir_path):
+        for file in files:
+            file_path = os.path.join(dir_path, file)
+            Jtr = load_Jtr(file_path)
+            ans[file]=cross_frames(Jtr)
+    return ans
+            
+if __name__ == "__main__":
+    args=parse_args()
+    d=cross_detector(args.output_path)
+    json.dump(d,open("./fit/output/cross_detection/{}.json".format(args.dataset_name),'w'))
\ No newline at end of file