search

MendelXu / SAN

Open-vocabulary Semantic Segmentation
https://mendelxu.github.io/SAN/
MIT License
295 stars 27 forks source link

About Visualization of Results #41

Closed hly1999 closed 4 months ago

hly1999 commented 8 months ago

I would like to ask where the visualization part of the results in the code is located.Thanks

MendelXu commented 7 months ago

You can find it here

hly1999 commented 7 months ago

Sorry，I want to know how to see the segmented image，it doesn't seem to be in the output folder.

MendelXu commented 7 months ago

Could you provide the command you used?

hly1999 commented 7 months ago

python train_net.py --config-file configs/san_clip_vit_res4_coco.yaml --num-gpus 4 OUTPUT_DIR ./output2/trained_vit_b16 I cannot find the segmented JPG image in the output folder.

MendelXu commented 7 months ago

It will not produce the segmented image directly. The results are stored in ./output2/trained_vit_b16/inference/*.json. If you want to visualize it with https://gist.github.com/MendelXu/052d65bd51c48887274fd985eb0f3ad1. Before using it, change line 11 to from san import *

hly1999 commented 7 months ago

OK，thank you for your response.

hly1999 commented 6 months ago

我在运行上面的可视化文件时遇到了这个问题 image AttributeError: Attribute 'stuff_dataset_id_to_contiguous_id' does not exist in the metadata of dataset 'pcontext_full_sem_seg_val'. Available keys are dict_keys(['name', 'stuff_classes', 'image_root', 'sem_seg_root', 'evaluator_type', 'ignore_label']). 请问这个是什么原因

MendelXu commented 4 months ago

refer to https://github.com/MendelXu/SAN/commit/3998b05867bb9879398727bf8478700ce3c005fb