Set geometry in geodataframe to another column fails : TypeError: Input must be valid geometry objects:

Question

I have CSV file which I read with geopandas. I use "read_file",define the crs , but seems like it doesn't know to choose the correct geometry column. I have tried to use "set_geometry" but then I get the typeerror:

TypeError: Input must be valid geometry objects:...

When I save teh csv as shapefile and then read it it does identify the geoemtry column. This is sample of my dataframe:

>>>name       GEOMETRY                                                                        
0   Greens  MultiPolygon (((-53.75361862037225791 -16.81549027035698884, -53.7434949680008458 -16.80934187415235925, -53.73225824390273431 -16.81517224986364667, -53.74868930272544532 -16.82322876902832931, -53.75160449058108725 -16.81893549236820107, -53.75361862037225791 -16.81549027035698884)))     
1  Tomas    MultiPolygon (((-53.76220054670636728 -16.80317298894741285, -53.75584013683950957 -16.81220477095834909, -53.76697085410650345 -16.8175475152465097, -53.77237720249333108 -16.81086908488630982, -53.77364928446670689 -16.80069242909933891, -53.76220054670636728 -16.80317298894741285)))
2  Lilians  MultiPolygon (((-53.76402937655674918 -16.78765299523402277, -53.75130855682304087 -16.78781200548069208, -53.7503544953430108 -16.80424306430340664, -53.76662654391905249 -16.79772364418987962, -53.76402937655674918 -16.78765299523402277)))

When I read it from csv:

shapes = gpd.read_file('./shapes_csv/polygons.csv')
shapes.crs = 'epsg:4326'

#checking the dtypes of columns: 
#we get new additional geometry column which is null:
shapes.dtypes
>>>
name          object
GEOMETRY      object
id            object
geometry    geometry


#trying to use set geoemtry in otder to get geoemtry values:

shapes=shapes.set_geometry('GEOMETRY')

>>>TypeError: Input must be valid geometry objects: MultiPolygon (((-53.75361862037225791 -16.81549027035698884, -53.7434949680008458 -16.80934187415235925, -53.73225824390273431 -16.81517224986364667, -53.74868930272544532 -16.82322876902832931, -53.75160449058108725 -16.81893549236820107, -53.75361862037225791 -16.81549027035698884)))

If I save the csv as shapefile in qgis and try to read it, then the geoemtry column is identified no problem:

shapes = gpd.read_file('./shapes_csv/polygons.shp')
shapes.dtypes
>>>
id            object
geometry    geometry
dtype: object

as you can see ,as shape no additionl column is added and I get the correct geometry column to be identified as geometry. My question is why is this hapenning? and how can I set the correct geometry column from geodataframe that is read from csv?

Answer 1

When you open a csv file, with GeoPandas it automatically adds a geometry field, empty here

gdf = gpd.read_file('myFile.csv')
print(gdf)       
          GEOMETRY                                     name     geometry  
0  MULTIPOLYGON (((-53.753618620372258 -16.815490...   Greens     None
1  MULTIPOLYGON (((-53.762200546706367 -16.803172...    Tomas     None
2  MULTIPOLYGON (((-53.764029376556749 -16.787652...  Lilians     None

The format of yours MultiPolygons is WKT (in GEOMETRY), so you have to convert it to shapely MultiPolygons (Converting a column of Polygons from string to GeoPandas geometry, How do I import a .csv file with a column of coordinates for polygons to a geodataframe?,Creating a GeoDataFrame from a DataFrame with coordinates,...)

from shapely.wkt import loads
gdf.geometry =  gdf['GEOMETRY'].apply(loads)