Set the geometry as the dataframes boundary, intersect the df with itself, select rows where ids are different:

df["id"] = df["id"].astype(int)
df.geometry = df.boundary
df = gpd.overlay(df1=df, df2=df, how="intersection")
df = df.loc[df["id_1"] < df["id_2"]]

Set the geometry to the dataframes boundary, intersect the df with itself, select rows where ids are different:

df["id"] = df["id"].astype(int)
df.geometry = df.boundary
df = gpd.overlay(df1=df, df2=df, how="intersection")
df = df.loc[df["id_1"] < df["id_2"]]

You can spatial join the df to itself by predicate touches, and intersect the resulting geometries:

#Save the geometry in a column
df["the_geom"] = df.geometry 
#Spatial join the df to itself, by predicate touches. So adjacent polygons are joined together
sj = df.sjoin(df, how="inner", predicate="touches")

#Polygon 2 is adjacent to 1, and 4 to 1:
# sj[["id_left","id_right"]].head(2)
#    id_left  id_right
# 0      2.0       1.0
# 3      4.0       1.0

# sj.columns
# ['id_left', 'geometry', 'the_geom_left', 'index_right', 'id_right',
#        'the_geom_right']

#Drop duplicate joins, like id 0-1 and id 1-0. I dont know if you want to keep them or not.
duplicate_mask = gpd.pd.DataFrame(np.sort(a=sj[["id_left","id_right"]].values, axis=-1)).duplicated().tolist()
sj = sj.loc[duplicate_mask]

#Intersect adjacent polygons, the polygons on the same row in the sj dataframe.
    #The result will be line and multilines

sj[["id_left","id_right"]]
sj["border_lines"] = sj.apply(lambda x: x.the_geom_left.intersection(x.the_geom_right), axis=1)
sj = sj.set_geometry("border_lines")#.explode()
sj = sj.drop(columns=["geometry", "the_geom_left", "the_geom_right", "index_right"])

I dont get your row count but you should be able to adjust the code to get what you want.

Set the geometry as the dataframes boundary, intersect the df with itself, select rows where ids are different:

df["id"] = df["id"].astype(int)
df.geometry = df.boundary
df = gpd.overlay(df1=df, df2=df, how="intersection")
df = df.loc[df["id_1"] < df["id_2"]]