A Persian Image Captioning model based on Vision Encoder Decoder Models of the transformers🤗.

Overview

Persian-Image-Captioning

Hugging Face Spaces

We fine-tuning the Vision Encoder Decoder Model for the task of image captioning on the coco-flickr-farsi dataset. The implementation of our model is in PyTorch with transformers library by Hugging Face( 🤗 ).

You can choose any pretrained vision model and any language model to use in the Vision Encoder Decoder model. Here we use ViT as the encoder, and ParsBERT (v2.0) as the decoder. The encoder and decoder are loaded separately via from_pretrained()function. Cross-attention layers are randomly initialized and added to the decoder.

You may refer to the Vision Encoder Decoder Model for more information.

How to use

You can generate caption of an image using this model using the code below:

import torch
import urllib
import PIL
import matplotlib.pyplot as plt
from transformers import ViTFeatureExtractor, AutoTokenizer, \
                         VisionEncoderDecoderModel

def show_img(image):
    # show image
    plt.axis("off")
    plt.imshow(image)
    
if torch.cuda.is_available():
    device = 'cuda'
else:
    device = 'cpu'


#pass the url of any image to generate a caption for it    
urllib.request.urlretrieve("https://images.unsplash.com/photo-1628191011227-522c7c3f0af9?ixlib=rb-1.2.1&ixid=MnwxMjA3fDB8MHxwaG90by1wYWdlfHx8fGVufDB8fHx8&auto=format&fit=crop&w=870&q=80", "sample.png")
image = PIL.Image.open("sample.png")


#Load the model you trained for inference 
model_checkpoint = 'MahsaShahidi/Persian-Image-Captioning'
model = VisionEncoderDecoderModel.from_pretrained(model_checkpoint).to(device)

feature_extractor = ViTFeatureExtractor.from_pretrained('google/vit-base-patch16-224-in21k')
tokenizer = AutoTokenizer.from_pretrained('HooshvareLab/bert-fa-base-uncased-clf-persiannews')

sample = feature_extractor(image, return_tensors="pt").pixel_values.to(device)
caption_ids = model.generate(sample, max_length = 30)[0]
caption_text = tokenizer.decode(caption_ids, skip_special_tokens=True)
print(caption_text)
show_img(image)

Inference

Following are the reslts of 3 captions generated on free stock photos after 2 epochs of training.

Image Caption
Generated Caption: زنی در آشپزخانه در حال اماده کردن غذا است.
Generated Caption: گروهی از مردم در حال پرواز بادبادک در یک زمین چمنزار.
Generated Caption: مردی در ماشین نشسته و به ماشین نگاه می کند.

Credits

A huge thanks to Kaggle for providing free access to GPU, and to the creators of Huggingface, ViT, and ParsBERT!

References

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

You might also like...
An Analysis Toolkit for Natural Language Generation (Translation, Captioning, Summarization, etc.)
An Analysis Toolkit for Natural Language Generation (Translation, Captioning, Summarization, etc.)

VizSeq is a Python toolkit for visual analysis on text generation tasks like machine translation, summarization, image captioning, speech translation

Python code for ICLR 2022 spotlight paper EViT: Expediting Vision Transformers via Token Reorganizations
Python code for ICLR 2022 spotlight paper EViT: Expediting Vision Transformers via Token Reorganizations

Expediting Vision Transformers via Token Reorganizations This repository contain

Implementation of the Hybrid Perception Block and Dual-Pruned Self-Attention block from the ITTR paper for Image to Image Translation using Transformers
Implementation of the Hybrid Perception Block and Dual-Pruned Self-Attention block from the ITTR paper for Image to Image Translation using Transformers

ITTR - Pytorch Implementation of the Hybrid Perception Block (HPB) and Dual-Pruned Self-Attention (DPSA) block from the ITTR paper for Image to Image

Yet Another Sequence Encoder - Encode sequences to vector of vector in python !

Yase Yet Another Sequence Encoder - encode sequences to vector of vectors in python ! Why Yase ? Yase enable you to encode any sequence which can be r

REST API for sentence tokenization and embedding using Multilingual Universal Sentence Encoder.

What is MUSE? MUSE stands for Multilingual Universal Sentence Encoder - multilingual extension (16 languages) of Universal Sentence Encoder (USE). MUS

ALIbaba's Collection of Encoder-decoders from MinD (Machine IntelligeNce of Damo) Lab

AliceMind AliceMind: ALIbaba's Collection of Encoder-decoders from MinD (Machine IntelligeNce of Damo) Lab This repository provides pre-trained encode

PyTorch implementation and pretrained models for XCiT models. See XCiT: Cross-Covariance Image Transformer
PyTorch implementation and pretrained models for XCiT models. See XCiT: Cross-Covariance Image Transformer

Cross-Covariance Image Transformer (XCiT) PyTorch implementation and pretrained models for XCiT models. See XCiT: Cross-Covariance Image Transformer L

simpleT5 is built on top of PyTorch-lightning⚡️ and Transformers🤗 that lets you quickly train your T5 models.
simpleT5 is built on top of PyTorch-lightning⚡️ and Transformers🤗 that lets you quickly train your T5 models.

Quickly train T5 models in just 3 lines of code + ONNX support simpleT5 is built on top of PyTorch-lightning ⚡️ and Transformers 🤗 that lets you quic

KoBART model on huggingface transformers

KoBART-Transformers SKT에서 공개한 KoBART를 편리하게 사용할 수 있게 transformers로 포팅하였습니다. Install (Optional) BartModel과 PreTrainedTokenizerFast를 이용하면 설치하실 필요 없습니다. p

Owner
Hamtech-ai
Hamtech-ai
A fast and lightweight python-based CTC beam search decoder for speech recognition.

pyctcdecode A fast and feature-rich CTC beam search decoder for speech recognition written in Python, providing n-gram (kenlm) language model support

Kensho 306 Nov 20, 2022
End-to-end image captioning with EfficientNet-b3 + LSTM with Attention

Image captioning End-to-end image captioning with EfficientNet-b3 + LSTM with Attention Model is seq2seq model. In the encoder pretrained EfficientNet

null 2 Feb 10, 2022
Beyond Masking: Demystifying Token-Based Pre-Training for Vision Transformers

beyond masking Beyond Masking: Demystifying Token-Based Pre-Training for Vision Transformers The code is coming Figure 1: Pipeline of token-based pre-

Yunjie Tian 23 Sep 27, 2022
Transformer-based Text Auto-encoder (T-TA) using TensorFlow 2.

T-TA (Transformer-based Text Auto-encoder) This repository contains codes for Transformer-based Text Auto-encoder (T-TA, paper: Fast and Accurate Deep

Jeong Ukjae 12 Apr 23, 2021
PyTorch Implementation of VAENAR-TTS: Variational Auto-Encoder based Non-AutoRegressive Text-to-Speech Synthesis.

VAENAR-TTS - PyTorch Implementation PyTorch Implementation of VAENAR-TTS: Variational Auto-Encoder based Non-AutoRegressive Text-to-Speech Synthesis.

Keon Lee 67 Nov 14, 2022
jel - Japanese Entity Linker - is Bi-encoder based entity linker for japanese.

jel: Japanese Entity Linker jel - Japanese Entity Linker - is Bi-encoder based entity linker for japanese. Usage Currently, link and question methods

izuna385 9 Nov 18, 2022
Yuqing Xie 2 Feb 17, 2022
An implementation of model parallel GPT-3-like models on GPUs, based on the DeepSpeed library. Designed to be able to train models in the hundreds of billions of parameters or larger.

GPT-NeoX An implementation of model parallel GPT-3-like models on GPUs, based on the DeepSpeed library. Designed to be able to train models in the hun

EleutherAI 2.7k Dec 3, 2022
Flexible interface for high-performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra.

Flexible interface for high performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra. What is Lightning Tran

Pytorch Lightning 575 Dec 2, 2022
An Analysis Toolkit for Natural Language Generation (Translation, Captioning, Summarization, etc.)

VizSeq is a Python toolkit for visual analysis on text generation tasks like machine translation, summarization, image captioning, speech translation

Facebook Research 409 Oct 28, 2022