#!/bin/python3

import math

import os

import random

import re

import sys

from statistics import median

"""

This solution hits timeout exceptions on test cases.

I think it's because computing the median of the trailing days can be done

faster than using Python's median function. Instead consider using heaps.

"""

def getTrailingExpenditures(dayNumber, dailyExpenditures, trailingDays):

return dailyExpenditures[dayNumber - trailingDays: dayNumber]

def getTrailingMedian(dayNumber, dailyExpenditures, trailingDays):

trailingExpenditures = getTrailingExpenditures(

dayNumber, dailyExpenditures, trailingDays)

return median(trailingExpenditures)

def activityNotificationOnDay(dayNumber, dailyExpenditures, trailingDays):

"""

Returns True if there an activity notification will be triggered on dayNumber

Returns False if no activity notification will be triggered on dayNumber

"""

todaySpend = dailyExpenditures[dayNumber]

trailingMedian = getTrailingMedian(

dayNumber, dailyExpenditures, trailingDays)

return todaySpend >= 2 * trailingMedian

def activityNotifications(dailyExpenditures, trailingDays):

activityNotifications = 0

for dayNumber in range(trailingDays, len(dailyExpenditures)):

if activityNotificationOnDay(

dayNumber,

dailyExpenditures,

trailingDays):

activityNotifications += 1

return activityNotifications

if __name__ == '__main__':

fptr = open(os.environ['OUTPUT_PATH'], 'w')

n, trailingDays = map(int, input().split())

dailyExpenditures = list(map(int, input().rstrip().split()))

result = activityNotifications(dailyExpenditures, trailingDays)

fptr.write(str(result) + '\n')

fptr.close()