diff --git a/bak/init_data.py b/bak/init_data.py
index 9590782..1d4ea10 100644
--- a/bak/init_data.py
+++ b/bak/init_data.py
@@ -25,3 +25,16 @@ class BatchDataRead(BatchDataBase):
         orm_mode = True
 
 
+class BatchDataCreate(BatchDataBase):
+    year: int | None = 2024
+    census_batch: str | None = ''  # 普查批次
+    id_code: str | None = ''  # 编号
+    precision: str | None = ''  # 精度
+    is_train: bool | None = True
+    is_validation: bool | None = False
+
+    ann_file: str | None = ''
+    ann_file_lbs: str | None = ''
+    img_prefix: str | None = ''
+    filter_empty_gt: bool | None = False
+    update_cache: bool | None = False
diff --git a/main.py b/main.py
index 62b25dc..2bd87f5 100644
--- a/main.py
+++ b/main.py
@@ -5,7 +5,7 @@ import pandas as pd
 import streamlit as st
 from loguru import logger
 
-from bak.init_data import BatchDataRead
+from bak.init_data import BatchDataCreate, BatchDataRead
 from db_utils import BatchData, session
 
 BASE_CKPT_DIR = "./bak"
@@ -56,22 +56,44 @@ def train():
     st.session_state.configs['evolve_r'] = st.session_state.evolve_r
 
 
-def update_handler():
+def table_update_handler():
+    logger.debug(f"{st.session_state.edited_info=}")
     edited_rows = st.session_state.edited_info.get('edited_rows')
+    added_rows = st.session_state.edited_info.get('added_rows')
+    deleted_rows = st.session_state.edited_info.get('deleted_rows')
+    # 更改列
     for id_, update_data in edited_rows.items():
         row_id = int(edited_df.loc[id_].id)
         row_db = session.query(BatchData).where(BatchData.id == row_id).first()
         logger.info(f"{row_id=}, {update_data=}")
-        logger.debug(BatchDataRead.from_orm(row_db))
         for field in update_data:
             setattr(row_db, field, update_data[field])
         session.commit()
 
+    # 增加列
+    for new_row_data in added_rows:
+        logger.debug(f"{new_row_data=}")
+
+    for deleted_row in deleted_rows:
+        row_id = int(edited_df.loc[deleted_row].id)
+        logger.debug(f"{row_id=}")
+        row_db = session.query(BatchData).where(BatchData.id == row_id).first()
+        session.delete(row_db)
+        session.commit()
+
 
 def update_config(*args, **kwargs):
     return None
 
 
+def create_new_row():
+    session.add(
+        BatchData(**BatchDataCreate().dict())
+    )
+    session.commit()
+    st.session_state.data_table = get_data_from_db()
+
+
 with left_col:
     edited_df = st.data_editor(
         df, key="edited_info",
@@ -79,7 +101,7 @@ with left_col:
         height=600,
         hide_index=True,
         use_container_width=True,
-        on_change=update_handler,
+        on_change=table_update_handler,
         column_order=(
             'id', 'year', 'census_batch', 'id_code', 'precision', 'is_train', 'is_validation', 'ann_file', 'ann_file_lbs', 'img_prefix',),
         column_config={
@@ -93,6 +115,7 @@ with left_col:
             'ann_file_lbs': st.column_config.Column("lbs_path", width='medium'),
             'img_prefix': st.column_config.Column("img_prefix", width='medium'),
         })
+    st.button("新建一行", on_click=create_new_row)
 
 with right_col:
     st.slider(label='evolve_r', key='evolve_r',