無法轉換yolov3-tiny.weights為yolov3-tiny.h5

2020/01/30 上午 08:55

電腦視覺深度學習討論版

OwenChiu

觀看數：3

回答數：2

收藏數：0

cvdl-1

cvdl-1-d40

執行以下程式

# 下載 tiny-yolov3 的網路權重，並且把權重轉換為 keras 能夠讀取的格式

tiny_yolo_model_path = "model_data/yolov3-tiny.h5"

tiny_yolo_anchor_file = "model_data/tiny_yolo_anchors.txt"

if not os.path.exists(tiny_yolo_model_path):

print("yolov3-tiny weights doesn't exist, downloading...")

print("Converting yolov3-tiny.weights to yolov3-tiny.h5...")

os.system("python convert.py yolov3-tiny.cfg yolov3-tiny.weights %s" % tiny_yolo_model_path)

if os.path.exists(tiny_yolo_model_path):

print("Done!")

else:

print("Strange, model doesn't exist, pleace check")

結果為

yolov3-tiny weights doesn't exist, downloading...

Converting yolov3-tiny.weights to yolov3-tiny.h5...

Strange, model doesn't exist, pleace check

無法轉換成功