「TypeError: slice indices must be integers or None or have an __index_

###前提・実現したいこと
Pythonで掲題のエラーが発生してしまい困っております。皆様のお知恵を拝借したく。何卒宜しくお願いします。

###発生している問題・エラーメッセージ

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-274-1ab25a298417> in <module>()
----> 1 mask_L,mask_R = initial_mask(lane_line_gb,window_size=40)
      2 
      3 plt.subplot(1,2,1)
      4 plt.imshow(mask_L,cmap='gray')
      5 plt.axis('off')

<ipython-input-273-f5ded2734466> in initial_mask(img, window_size)
     12     mov_filtsize = img_size[1]/50.
     13     mean_lane = np.mean(img[img_size[1]//2:,:],axis=0)
---> 14     mean_lane = moving_average(mean_lane,mov_filtsize)
     15 
     16     #find peak indexes from histgram

<ipython-input-273-f5ded2734466> in moving_average(a, n)
      4     # Moving average
      5     ret = np.cumsum(a, dtype=float)
----> 6     ret[n:] = ret[n:] - ret[:-n]
      7     return ret[n - 1:] / n
      8 

TypeError: slice indices must be integers or None or have an __index__ method

###該当のソースコード（Python3.5）

Python3.5
1from scipy.signal import find_peaks_cwt
2
3def moving_average(a, n=3):
4    # Moving average
5    ret = np.cumsum(a, dtype=float)
6    ret[n:] = ret[n:] - ret[:-n]
7    return ret[n - 1:] // n
8
9def initial_mask(img,window_size):
10    img = gaussian_blur(img,5)
11    img_size = np.shape(img)
12    mov_filtsize = img_size[1]/50.
13    mean_lane = np.mean(img[img_size[1]//2:,:],axis=0)
14    mean_lane = moving_average(mean_lane,mov_filtsize)
15    
16    #find peak indexes from histgram
17    indexes = find_peaks_cwt(mean_lane,[100],max_distances=[800])
18    
19    val_ind = np.array([mean_lane[indexes[i]] for i in range(len(indexes)) ])
20    ind_sorted = np.argsort(-val_ind)
21
22    ind_peakR = indexes[ind_sorted[0]]
23    ind_peakL = indexes[ind_sorted[1]]
24    
25    # exchange indexes if under condition
26    if ind_peakR<ind_peakL:
27        
28        ind_peakR,ind_peakL = ind_peakL,ind_peakR
29
30    # take windows from each index
31    ind_min_L = ind_peakL-window_size
32    ind_max_L = ind_peakL+window_size
33
34    ind_min_R = ind_peakR-window_size
35    ind_max_R = ind_peakR+window_size
36
37    mask_L = np.zeros_like(img)
38    mask_R = np.zeros_like(img)
39
40    ind_peakR_prev = ind_peakR
41    ind_peakL_prev = ind_peakL
42    
43    # Split image into 11 parts and compute histogram on each part
44    
45    for i in range(11):
46        img_y1 = img_size[0]-img_size[0]*i/11
47        img_y2 = img_size[0]-img_size[0]*(i+1)/11
48    
49        mean_lane_y = np.mean(img[img_y2:img_y1,:],axis=0)
50        mean_lane_y = moving_average(mean_lane_y,mov_filtsize)
51        indexes = find_peaks_cwt(mean_lane_y,[100], max_distances=[800])
52        
53        # if the indexes are more than 2 (both side peak L,R)
54        if len(indexes)>1.5:
55            val_ind = np.array([mean_lane[indexes[i]] for i in range(len(indexes)) ])
56            ind_sorted = np.argsort(-val_ind)
57
58            ind_peakR = indexes[ind_sorted[0]]
59            ind_peakL = indexes[ind_sorted[1]]
60            if ind_peakR<ind_peakL:
61                ind_peakR,ind_peakL = ind_peakL,ind_peakR
62            
63        else:
64        # if one peak is found
65            if len(indexes)==1:
66                #found right lane 
67                if (np.abs(indexes[0]-ind_peakR_prev)<np.abs(indexes[0]-ind_peakL_prev)):
68                    ind_peakR = indexes[0]
69                    ind_peakL = ind_peakL_prev
70                # found left lane
71                else:
72                    ind_peakL = indexes[0]
73                    ind_peakR = ind_peakR_prev
74            # If no pixels are found, use previous ones. 
75            else:
76                ind_peakL = ind_peakL_prev
77                ind_peakR = ind_peakR_prev
78            
79            
80        # If new center is more than 80pixels away, use previous
81        # Outlier rejection
82        if np.abs(ind_peakL-ind_peakL_prev)>=80:
83            ind_peakL = ind_peakL_prev
84
85        if np.abs(ind_peakR-ind_peakR_prev)>=80:
86            ind_peakR = ind_peakR_prev
87            
88    
89            
90        mask_L[img_y2:img_y1,ind_peakL-window_size:ind_peakL+window_size] = 1.
91        mask_R[img_y2:img_y1,ind_peakR-window_size:ind_peakR+window_size] = 1.
92        
93        ind_peakL_prev = ind_peakL
94        ind_peakR_prev = ind_peakR
95        
96    return mask_L,mask_R
97    
98

Python3.5
1mask_L,mask_R = initial_mask(lane_line_gb,window_size=40)
2
3plt.subplot(1,2,1)
4plt.imshow(mask_L,cmap='gray')
5plt.axis('off')
6plt.title('Left lane')
7plt.subplot(1,2,2)
8plt.imshow(mask_R,cmap='gray')
9plt.axis('off')
10plt.title('Right lane');
11    
12

###試したこと
１．Stackoverflowを参考にしようとしましたが、解決方法が分かりませんでした。
URL:https://stackoverflow.com/questions/20733156/slice-indices-must-be-integers-or-none-or-have-index-method

###補足情報(言語/FW/ツール等のバージョンなど)
Python 3.5.2

                                                 －以上－

行動規範の内容に同意します

回答1件

ベストアンサー

initial_mask内の以下の部分を書き換えると、現状のエラーは消えるかもしれません。

Python
1def initial_mask(img,window_size):
2    img = gaussian_blur(img,5)
3    img_size = np.shape(img)
4    mov_filtsize = int(img_size[1]/50.)   # ← ここ
5
6    ...

参考までに、同じエラーが出るコードを。

Python
1>>> ringo = 'apple'
2>>> ringo[2:]
3'ple'
4>>> ringo[2.:]
5Traceback (most recent call last):
6  File "<stdin>", line 1, in <module>
7TypeError: slice indices must be integers or None or have an __index__ method
8>>> ringo[int(2.):]
9'ple'