Spaces:
Running
Running
| import os | |
| import torch | |
| import numpy as np | |
| from fairseq import utils,tasks | |
| from utils.checkpoint_utils import load_model_ensemble_and_task | |
| from models.polyformer import PolyFormerModel | |
| import cv2 | |
| import torch | |
| import numpy as np | |
| from fairseq import utils, tasks | |
| from fairseq import checkpoint_utils | |
| from utils.eval_utils import eval_step | |
| from tasks.refcoco import RefcocoTask | |
| from models.polyformer import PolyFormerModel | |
| from PIL import Image | |
| from torchvision import transforms | |
| import cv2 | |
| import gradio as gr | |
| import math | |
| from io import BytesIO | |
| import base64 | |
| import re | |
| from demo import visual_grounding | |
| title = "PolyFormer-Visual_Grounding" | |
| description = "Gradio Demo for PolyFormer-Visual_Grounding. Upload your own image or click any one of the examples, " \ | |
| "and write a description about a certain object. " \ | |
| "Then click \"Submit\" and wait for the result of grounding. For help or to provide feedback, please contact: Hui Ding (@huidin)" | |
| article = "<p style='text-align: center'><a href='https://arxiv.org/pdf/2302.07387.pdf' target='_blank'>PolyFormer CVPR2023" \ | |
| "</a></p> " | |
| # examples = [['A bear astronaut in the space.jpeg', 'a bear astronaut in the space'], | |
| # ['A unicorn doing computer vision research.jpeg', 'a unicorn doing computer vision research'], | |
| # ['pig.jpeg', 'a pig robot preparing a delicious meal'], | |
| # ['otta.png', 'a gentleman otter in a 19th century portrait'], | |
| # ['pikachu.jpeg', 'a pikachu fine-dining with a view to the Eiffel Tower'], | |
| # ['A small cabin on top of a snowy mountain in the style of Disney artstation.jpeg', 'a small cabin on top of a snowy mountain in the style of Disney artstation'], | |
| # | |
| # ] | |
| examples = [] | |
| io = gr.Interface(fn=visual_grounding, inputs=[gr.inputs.Image(type='pil'), "textbox"], | |
| outputs=[gr.outputs.Image(label="output", type='numpy'), gr.outputs.Image(label="predicted mask", type='numpy')], | |
| title=title, description=description, article=article, examples=examples, | |
| allow_flagging=False, allow_screenshot=False) | |
| # io.launch(cache_examples=True) | |
| io.launch() | |