2026最新！8个一键生成论文工具测评：专科生毕业论文+开题报告写作全攻略

说明及各模型下载https://github.com/opencv/opencv_contrib/tree/master/modules/dnn_superres结论时髦归时髦我在虚拟机中测试性能非常之差。

而且更可笑的是必须是整数这基本没有应用场景。

代码import cv2 from cv2 import dnn_superres import time def super_resolution(input_path, model_path, model_name, scale): # 初始化 sr dnn_superres.DnnSuperResImpl_create() print(model_path) # 读取模型 sr.readModel(model_path) sr.setModel(model_name, scale) # 读取图像 img cv

imread(input_path) if img is None: print(无法读取图像) return # 记录时间 start_time time.time() for index in range(count): # 超分辨率重建 result sr.upsample(img) print(%s cost time: %.2f ms % (model_path, (time.time()-start_time)*1000/count)) return result models[ [EDSR_x

pb, 2, edsr], [ESPCN_x

pb, 2, espcn], [FSRCNN-small_x

pb, 2, fsrcnn], [FSRCNN_x

pb, 2, fsrcnn], [LapSRN_x

pb, 2, lapsrn], [EDSR_x

pb, 3, edsr], [ESPCN_x

pb, 3, espcn], [FSRCNN-small_x

pb, 3, fsrcnn], [FSRCNN_x

pb, 3, fsrcnn], [LapSRN_x

pb, 4, lapsrn], [EDSR_x

pb, 4, edsr], [ESPCN_x

pb, 4, espcn], [FSRCNN-small_x

pb, 4, fsrcnn], [FSRCNN_x

pb, 4, fsrcnn], [LapSRN_x

pb, 8, lapsrn] ] test_filelotus-300x

jpg test_filelotus-600x

jpg count10 scale2 for model in models: super_resolution( test_file, models/model[0], model[2], model[1])

17C网页官方入口-17C网页官方入口应用