NDVI shape file creator from Landsat images, using Google earth engine and Python API

ndvi_gee_shapefile.py

import numpy as np
import fiona
from shapely import geometry
import ee
ee.Initialize()
from gee_library import *
from rasterio.features import shapes
import geopandas as gp
import ntpath
import glob
import json
import pandas as pd


shapefiles=glob.glob('*.shp')


def path_leaf(path):
    head, tail = ntpath.split(path)
    return tail or ntpath.basename(head)

collection = ee.ImageCollection('LANDSAT/LC8_L1T_TOA_FMASK')

wer=[]
for shpfiles in shapefiles:
    with fiona.open(shpfiles, "r") as shpfile:
        features = [feature["geometry"] for feature in shpfile]
    filename=(path_leaf(shpfiles)).split('.')[0]
    aa=geometry.Polygon(features[0]['coordinates'][0])
    bb=aa.bounds
    ll=(bb[0],bb[1])
    ur=(bb[2],bb[3])
    nps_bounds = bound_geometry(ll,ur)
    print filename
    #print(nps_bounds)
    grid_collection = collection.filterBounds(nps_bounds)
    tr_cr = grid_collection.filter(ee.Filter.date('2013-01-01', '2016-01-01'))
    cc_cr = tr_cr.filter(ee.Filter.lt('CLOUD_COVER', 10))
    aa=cc_cr.getInfo()
    we=[]
    for aas in aa['features']:
        we.append(aas['properties']['DATE_ACQUIRED'])
    print "there are "+str(len(we))+" images for "+filename 
    if len(we) >0:
        print "hello"
        tiles,img_affine = img_at_region(geCollection=cc_cr,resolution=30,bands=['B4', 'B5','fmask'],geo_bounds=nps_bounds)
        ndvi=(tiles['B5']-tiles['B4'])/(tiles['B5']+tiles['B4'])
        a_ndvi = np.where(ndvi<0.5,0,1)
        fmask=np.where(tiles['fmask']<=1,0,1)
        b_ndvi=a_ndvi+fmask
        c_ndvi=b_ndvi.astype(np.uint8)
        unique, counts = np.unique(c_ndvi, return_counts=True)
        dattif=dict(zip(unique, counts))
        dattif['date']=we[0]
        dattif['gridname']=filename
        wer.append(dattif)
        mask = None
        resultsU = ({'properties': {'raster_val': v}, 'geometry': s} for i, (s, v) in enumerate(shapes(c_ndvi, mask=mask, transform=img_affine)))
        geomsU = list(resultsU)
        dbU  = gp.GeoDataFrame.from_features(geomsU)
        dbU1=dbU[dbU['raster_val']==1]
        dbU1.to_file(filename+'.shp', driver='ESRI Shapefile')
    else: 
        print "no suitable image for "+filename
    grid_collection =[]
    tr_cr =[]
    cc_cr =[]
    #print filename


wer1=pd.DataFrame(wer)
wer1.to_csv('/home/sunbird/earthengine/ge_stats.csv')

Though I am not a python expert, the logical streams seems great. Wonderful job