SSD-Tensorflow的版本为，https://github.com/balancap/SSD-Tensorflow

引用GitHub上一个人的做法，https://github.com/balancap/SSD-Tensorflow/issues/154

训练完模型，想测试下其mAP的代码

DATASET_DIR=tfrecords 

EVAL_DIR=logs/ 

CHECKPOINT_PATH=./logs/VGG_VOC0712_SSD_300x300_ft_iter_120000.ckpt 

python3 eval_ssd_network.py \ 

    --eval_dir=${EVAL_DIR} \ 

	--dataset_dir=${DATASET_DIR} \ 

	--dataset_name=pascalvoc_2007 \ 

	--dataset_split_name=test \ 

	--model_name=ssd_300_vgg \ 

	--checkpoint_path=${CHECKPOINT_PATH} \ 

	--batch_size=1

出现错误
TypeError: Can not convert a tuple into a Tensor or Operation
解决方法为在eval_ssd_network.py文件中添加下面一个函数：

def flatten(x):

    result = []

    for el in x:

         if isinstance(el, tuple):

               result.extend(flatten(el))

         else:

               result.append(el)

    return result

然后修改eval_ssd_network.py中两行代码

将    eval_op=list(names_to_updates.values())

 

改为  eval_op=flatten(list(names_to_updates.values()))

 

注意：共有两行！