tasks/img_cap.py

# --------------------------------------------------------
# X-Decoder -- Generalized Decoding for Pixel, Image, and Language
# Copyright (c) 2022 Microsoft
# Licensed under The MIT License [see LICENSE for details]
# Written by Xueyan Zou (xueyan@cs.wisc.edu)
# --------------------------------------------------------

import cv2
import torch
import numpy as np
from PIL import Image
from torchvision import transforms


t = []
t.append(transforms.Resize(224, interpolation=Image.BICUBIC))
transform = transforms.Compose(t)

t = []
t.append(transforms.Resize(512, interpolation=Image.BICUBIC))
transform_v = transforms.Compose(t)

def image_captioning(model, image, texts, inpainting_text, *args, **kwargs):
    with torch.no_grad():
        image_ori = transform_v(image)
        width = image_ori.size[0]
        height = image_ori.size[1]
        image_ori = np.asarray(image_ori)

        image = transform(image)
        image = np.asarray(image)
        images = torch.from_numpy(image.copy()).permute(2,0,1).cuda()

        batch_inputs = [{'image': images, 'height': height, 'width': width, 'image_id': 0}]
        outputs = model.model.evaluate_captioning(batch_inputs)
        text = outputs[-1]['captioning_text']

        image_ori = image_ori.copy()
        cv2.rectangle(image_ori, (0, height-60), (width, height), (0,0,0), -1)
        font                   = cv2.FONT_HERSHEY_DUPLEX
        fontScale              = 1.2
        thickness              = 2
        lineType               = 2
        bottomLeftCornerOfText = (10, height-20)
        fontColor              = [255,255,255]
        cv2.putText(image_ori, text,
            bottomLeftCornerOfText,
            font, 
            fontScale,
            fontColor,
            thickness,
            lineType)
    torch.cuda.empty_cache()
    return Image.fromarray(image_ori), text, None