SegNetで重みが読み込めない

image-segmentation-kerasという以下1.のHPのコードを試しで動かしているのですが､
フォルダにある画像全部を予測したい場合の方法を探していると
同じHPのチュートリアル(2.HP)に記載されていました｡
しかし､そのコードを使うとAssertionError: Checkpoint not foundとの記載が出ます｡
エラーが出ない対処法をご存じの方教えていただければ助かります｡
よろしくお願いします｡

環境は以下です｡
WIN10
Anaconda
Python 3.6
keras 2.3.1
tensorflow 2.2.0

HPのURL
1.https://github.com/divamgupta/image-segmentation-keras
2.https://divamgupta.com/image-segmentation/2019/06/06/deep-learning-semantic-segmentation-keras.html

訓練のコード
from keras_segmentation.models.segnet import segnet
model = segnet(n_classes=2 , input_height=416, input_width=608 )
model.train(
train_images = "dataset2/images_prepped_train/",
train_annotations = "dataset2/annotations_prepped_train/",
checkpoints_path = "./tmp/segnet" , epochs=5)

予測コード(フォルダ内画像)
from keras_segmentation.predict import predict_multiple
predict_multiple(
inp_dir="dataset2/images_prepped_test/",
out_dir="outputs/",
checkpoints_path="./tmp/segnet",
)

暫定処置
import glob
predictData_input_path=glob.glob("dataset2/images_prepped_test/*")
for i in predictData_input_path:
name = os.path.split(i)[1]
out = model.predict_segmentation(
inp="dataset2/images_prepped_test/"+name,
out_fname="./tmp/"+name
)

エラー内容

AssertionError Traceback (most recent call last)
<ipython-input-4-7754888802dd> in <module>
5 inp_dir="dataset2/images_prepped_test/",
6 out_dir="outputs/",
----> 7 checkpoints_path="./tmp/segnet",
8 )

~\anaconda3\envs\GPU\lib\site-packages\keras_segmentation\predict.py in predict_multiple(model, inps, inp_dir, out_dir, checkpoints_path, overlay_img, class_names, show_legends, colors, prediction_width, prediction_height)
172
173 if model is None and (checkpoints_path is not None):
--> 174 model = model_from_checkpoint_path(checkpoints_path)
175
176 if inps is None and (inp_dir is not None):

~\anaconda3\envs\GPU\lib\site-packages\keras_segmentation\predict.py in model_from_checkpoint_path(checkpoints_path)
26 open(checkpoints_path+"_config.json", "r").read())
27 latest_weights = find_latest_checkpoint(checkpoints_path)
---> 28 assert (latest_weights is not None), "Checkpoint not found."
29 model = model_from_name[model_config['model_class']](
30 model_config['n_classes'], input_height=model_config['input_height'],

AssertionError: Checkpoint not found.