我们针对特定任务提供了一些脚本,可以对单张图像进行推理。
您可以使用以下命令,输入一张测试图像以及缺损部位的遮罩图像,实现对测试图像的补全。
python demo/inpainting_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${MASKED_IMAGE_FILE} \
${MASK_FILE} \
${SAVE_FILE} \
[--imshow] \
[--device ${GPU_ID}]
如果指定了 --imshow ,演示程序将使用 opencv 显示图像。例子:
python demo/inpainting_demo.py \
configs/inpainting/global_local/gl_256x256_8x12_celeba.py \
https://download.openmmlab.com/mmediting/inpainting/global_local/gl_256x256_8x12_celeba_20200619-5af0493f.pth \
tests/data/image/celeba_test.png \
tests/data/image/bbox_mask.png \
tests/data/pred/inpainting_celeba.png
补全结果将保存在 tests/data/pred/inpainting_celeba.png
中。
您可以使用以下命令,输入一张测试图像以及对应的三元图(trimap),实现对测试图像的抠图。
python demo/matting_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${IMAGE_FILE} \
${TRIMAP_FILE} \
${SAVE_FILE} \
[--imshow] \
[--device ${GPU_ID}]
如果指定了 --imshow ,演示程序将使用 opencv 显示图像。例子:
python demo/matting_demo.py \
configs/mattors/dim/dim_stage3_v16_pln_1x1_1000k_comp1k.py \
work_dirs/dim_stage3/latest.pth \
tests/data/merged/GT05.jpg \
tests/data/trimap/GT05.png \
tests/data/pred/GT05.png
预测的 alpha 遮罩将保存在 tests/data/pred/GT05.png
中。
您可以使用以下命令来测试要恢复的图像。
python demo/restoration_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${IMAGE_FILE} \
${SAVE_FILE} \
[--imshow] \
[--device ${GPU_ID}] \
[--ref-path ${REF_PATH}]
如果指定了 --imshow
,演示程序将使用 opencv 显示图像。例子:
python demo/restoration_demo.py \
configs/restorers/esrgan/esrgan_x4c64b23g32_g1_400k_div2k.py \
work_dirs/esrgan_x4c64b23g32_g1_400k_div2k/latest.pth \
tests/data/lq/baboon_x4.png \
demo/demo_out_baboon.png
您可以通过提供 --ref-path
参数来测试基于参考的超分辨率算法。例子:
python demo/restoration_demo.py \
configs/restorers/ttsr/ttsr-gan_x4_c64b16_g1_500k_CUFED.py \
https://download.openmmlab.com/mmediting/restorers/ttsr/ttsr-gan_x4_c64b16_g1_500k_CUFED_20210626-2ab28ca0.pth \
tests/data/test_multiple_gt/sequence_1/00000000.png \
work_dirs/demo_out.png \
--ref-path tests/data/test_multiple_gt/sequence_1/00000001.png
您可以使用以下命令来测试要恢复的人脸图像。
python demo/restoration_face_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${IMAGE_FILE} \
${SAVE_FILE} \
[--upscale-factor] \
[--face-size] \
[--imshow] \
[--device ${GPU_ID}]
如果指定了 --imshow ,演示程序将使用 opencv 显示图像。例子:
python demo/restoration_face_demo.py \
configs/restorers/glean/glean_in128out1024_4x2_300k_ffhq_celebahq.py \
https://download.openmmlab.com/mmediting/restorers/glean/glean_in128out1024_4x2_300k_ffhq_celebahq_20210812-acbcb04f.pth \
tests/data/face/000001.png \
results/000001.png \
--upscale-factor 4
您可以使用以下命令来测试视频以进行恢复。
python demo/restoration_video_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${INPUT_DIR} \
${OUTPUT_DIR} \
[--window-size=${WINDOW_SIZE}] \
[--device ${GPU_ID}]
它同时支持滑动窗口框架和循环框架。 例子:
EDVR:
python demo/restoration_video_demo.py \
./configs/restorers/edvr/edvrm_wotsa_x4_g8_600k_reds.py \
https://download.openmmlab.com/mmediting/restorers/edvr/edvrm_wotsa_x4_8x4_600k_reds_20200522-0570e567.pth \
data/Vid4/BIx4/calendar/ \
./output \
--window-size=5
BasicVSR:
python demo/restoration_video_demo.py \
./configs/restorers/basicvsr/basicvsr_reds4.py \
https://download.openmmlab.com/mmediting/restorers/basicvsr/basicvsr_reds4_20120409-0e599677.pth \
data/Vid4/BIx4/calendar/ \
./output
复原的视频将保存在 output/
中。
您可以使用以下命令来测试视频插帧。
python demo/video_interpolation_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${INPUT_DIR} \
${OUTPUT_DIR} \
[--fps-multiplier ${FPS_MULTIPLIER}] \
[--fps ${FPS}]
${INPUT_DIR}
和 ${OUTPUT_DIR}
可以是视频文件路径或存放一系列有序图像的文件夹。
若 ${OUTPUT_DIR}
是视频文件地址,其帧率可由输入视频帧率和 fps_multiplier
共同决定,也可由 fps
直接给定(其中前者优先级更高)。例子:
由输入视频帧率和 fps_multiplier
共同决定输出视频的帧率:
python demo/video_interpolation_demo.py \
configs/video_interpolators/cain/cain_b5_320k_vimeo-triplet.py \
https://download.openmmlab.com/mmediting/video_interpolators/cain/cain_b5_320k_vimeo-triple_20220117-647f3de2.pth \
tests/data/test_inference.mp4 \
tests/data/test_inference_vfi_out.mp4 \
--fps-multiplier 2.0
由 fps
直接给定输出视频的帧率:
python demo/video_interpolation_demo.py \
configs/video_interpolators/cain/cain_b5_320k_vimeo-triplet.py \
https://download.openmmlab.com/mmediting/video_interpolators/cain/cain_b5_320k_vimeo-triple_20220117-647f3de2.pth \
tests/data/test_inference.mp4 \
tests/data/test_inference_vfi_out.mp4 \
--fps 60.0
python demo/generation_demo.py \
${CONFIG_FILE} \
${CHECKPOINT_FILE} \
${IMAGE_FILE} \
${SAVE_FILE} \
[--unpaired-path ${UNPAIRED_IMAGE_FILE}] \
[--imshow] \
[--device ${GPU_ID}]
如果指定了 --unpaired-path
(用于 CycleGAN),模型将执行未配对的图像到图像的转换。 如果指定了 --imshow
,演示也将使用opencv显示图像。 例子:
针对配对数据:
python demo/generation_demo.py \
configs/example_config.py \
work_dirs/example_exp/example_model_20200202.pth \
demo/demo.jpg \
demo/demo_out.jpg
针对未配对数据(用 opencv 显示图像):
python demo/generation_demo.py 、
configs/example_config.py \
work_dirs/example_exp/example_model_20200202.pth \
demo/demo.jpg \
demo/demo_out.jpg \
--unpaired-path demo/demo_unpaired.jpg \
--imshow
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》