Keras报错 ‘ValueError: 'pool5' is not in list’

很长的一个project，在keras下实现VGG16。

这是报错的整个代码段：

for roi, roi_context in zip(rois, rois_context):                      
    ins = [im_in, dmap_in, np.array([roi]), np.array([roi_context])]  
    print("Testing ROI {c}")                                          
    subtimer.tic()                                                    
    blobs_out = model.predict(ins)                                    
    subtimer.toc()                                                    
    print("Storing Results")                                          
    print(layer_names)                                                
    post_roi_layers = set(layer_names[layer_names.index("pool5"):])   
    for name, val in zip(layer_names, blobs_out):                     
        if name not in outs:                                          
            outs[name] = val                                          
        else:                                                         
            if name in post_roi_layers:                               
                outs[name] = np.concatenate([outs[name], val])        
    c += 1                                                            

报错信息：

 Loading Test Data
data is loaded from roidb_test_19_smol.pkl
Number of Images to test: 10
Testing ROI {c}
Storing Results
['cls_score', 'bbox_pred_3d']
Traceback (most recent call last):
  File "/Users/xijiejiao/Amodal3Det_TF/tfmodel/main.py", line 6, in <module>
    results = test_main.test_tf_implementation(cache_file="roidb_test_19_smol.pkl", weights_path="rgbd_det_iter_40000.h5")
  File "/Users/xijiejiao/Amodal3Det_TF/tfmodel/test_main.py", line 36, in test_tf_implementation
    results = test.test_net(tf_model, roidb)
  File "/Users/xijiejiao/Amodal3Det_TF/tfmodel/test.py", line 324, in test_net
    im_detect_3d(net, im, dmap, test['boxes'], test['boxes_3d'], test['rois_context'])
  File "/Users/xijiejiao/Amodal3Det_TF/tfmodel/test.py", line 200, in im_detect_3d
    post_roi_layers = set(layer_names[layer_names.index("pool5"):])
ValueError: 'pool5' is not in list

你的 layer_names 只有['cls_score', 'bbox_pred_3d']两个元素，没有你要的pool5，仔细核对一下模型的加载对不对