This gist includes a function called stylable_groupby_boxplot() that receives a pandas dataframe object and the column you want to groupby and returns a dictionary that includes all the boxplot's parts just like the standard matplotlib boxplot function does.

Stylable Pandas Groupby Boxplots

from numpy.random import rand
import matplotlib.pyplot as plt
import pandas as pd

# 2 columns produces an array of 2 matplotlib.axes.AxesSubplot objects
df2 = pd.DataFrame(rand(10,2), columns=['Col1', 'Col2'] )
df2['X'] = pd.Series(['A','B','A','B','A','B','A','B','A','B'])

#1 column produces a single matplotlib.axes.AxesSubplot object
df1 = pd.DataFrame(rand(10), columns=['Col1'] )
df1['X'] = pd.Series(['A','B','A','B','A','B','A','B','A','B'])

def stylable_groupby_boxplot(df, by):
    '''
    If you plot only one column, boxplot returns a single AxesSubplot object.
    If there are several columns, boxplot returns an array of several AxesSubplot objects.
    '''
    bp = df.boxplot(by=by, grid=False)
    bptype = str(type(bp))
    if bptype == "<class 'matplotlib.axes.AxesSubplot'>":
        cl = bp.get_children()
        cl=[item for item in cl if isinstance(item, matplotlib.lines.Line2D)]
        bpdict = {}
        groups = df.groupby(by).groups.keys()
        for i in range(len(groups)):
            bpdict[groups[i]] = {'boxes': [], 'caps': [], 'fliers': [], 'medians': [], 'whiskers': []}
            bpdict[groups[i]]['boxes'] = [cl[4+8*i]]
            bpdict[groups[i]]['caps'] = [cl[2+8*i], cl[3+8*i]]
            bpdict[groups[i]]['fliers'] = [cl[6+8*i], cl[7+8*i]]
            bpdict[groups[i]]['medians'] = [cl[5+8*i]]
            bpdict[groups[i]]['whiskers'] = [cl[0+8*i], cl[1+8*i]]
    else:
        bpdict = {}
        groups = df.groupby(by).groups.keys()
        keys = range(len(bp))
        for i in keys:
            bpdict[keys[i]] = {}
            cl = bp[i].get_children()
            cl=[item for item in cl if isinstance(item, matplotlib.lines.Line2D)]
            for j in range(len(groups)):
                bpdict[keys[i]][groups[j]] = {'boxes': [], 'caps': [], 'fliers': [], 'medians': [], 'whiskers': []}
                bpdict[keys[i]][groups[j]]['boxes'] = [cl[4+8*j]]
                bpdict[keys[i]][groups[j]]['caps'] = [cl[2+8*j], cl[3+8*j]]
                bpdict[keys[i]][groups[j]]['fliers'] = [cl[6+8*j], cl[7+8*j]]
                bpdict[keys[i]][groups[j]]['medians'] = [cl[5+8*j]]
                bpdict[keys[i]][groups[j]]['whiskers'] = [cl[0+8*j], cl[1+8*j]]
    return bpdict

bp2 = stylable_groupby_boxplot(df2, by="X")
bp1 = stylable_groupby_boxplot(df1, by="X")


# 2 column styling
plt.suptitle("")
plt.setp(bp2[0]['A']['boxes'], color='blue')
plt.setp(bp2[0]['A']['medians'], color='red')
plt.setp(bp2[0]['A']['whiskers'], color='blue')
plt.setp(bp2[0]['A']['fliers'], color='blue')
plt.setp(bp2[0]['A']['caps'], color='blue')
plt.setp(bp2[0]['B']['boxes'], color='red')
plt.setp(bp2[0]['B']['medians'], color='blue')
plt.setp(bp2[0]['B']['whiskers'], color='red')
plt.setp(bp2[0]['B']['fliers'], color='red')
plt.setp(bp2[0]['B']['caps'], color='red')
plt.setp(bp2[1]['A']['boxes'], color='green')
plt.setp(bp2[1]['A']['medians'], color='purple')
plt.setp(bp2[1]['A']['whiskers'], color='green')
plt.setp(bp2[1]['A']['fliers'], color='green')
plt.setp(bp2[1]['A']['caps'], color='green')
plt.setp(bp2[1]['B']['boxes'], color='purple')
plt.setp(bp2[1]['B']['medians'], color='green')
plt.setp(bp2[1]['B']['whiskers'], color='purple')
plt.setp(bp2[1]['B']['fliers'], color='purple')
plt.setp(bp2[1]['B']['caps'], color='purple')

# 1 column styling
plt.suptitle("")
plt.setp(bp1['A']['boxes'], color='blue')
plt.setp(bp1['A']['medians'], color='red')
plt.setp(bp1['A']['whiskers'], color='blue')
plt.setp(bp1['A']['fliers'], color='blue')
plt.setp(bp1['A']['caps'], color='blue')
plt.setp(bp1['B']['boxes'], color='red')
plt.setp(bp1['B']['medians'], color='blue')
plt.setp(bp1['B']['whiskers'], color='red')
plt.setp(bp1['B']['fliers'], color='red')
plt.setp(bp1['B']['caps'], color='red')