Seach for pattern in dataframes

I have got a huge csv file which contains data about a bicycle being driven. So I got a time column in seconds and a Speed column. I would like to check for specific pattern in the data, in order to conclude what happend on the road.

For example driving torwards a traffic light:

I got this so far:

import pandas as pd

df = pd.read_csv('.csv', usecols = ['time', 'speed']) 
df['accelerating'] = df['speed'].diff() > 0

And I want something like this:

  df_traffic_light = df.loc[df['speed'] < 15 & accelerating == False #driving torwards the traffic light; 
    & df.loc[df['speed']< 1 #getting really slow or Standing still;
    & df.loc[df['speed']  > 5 & accelerating == True #for light switched to green and starting again 

Expected Output:

         time     speed  acceleration
0   5.000         14.0          false
1   7.056         12.0          false
2   10.097         8.0          false
3   12.131         1.0          false
4   14.165         0.0          false
5   16.201         0.0          false
6   18.236         2.0          true 
7   20.267         4.0          true

I tried it with dataframe.rolling but didnt work out quite well. Any ideas how I solve this?