Hugging Face's logo

deepklarity
/
poster2plot

Image Classification
Transformers
PyTorch
English
vision-encoder-decoder
image-captioning
Inference Endpoints
Model card Files Community
1
poster2plot
File size: 2,456 Bytes 
47962d5
 
 
 
 
 
 
 
 
 
 
 
749bdc7
 
47962d5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
 
---
language: en
tags:
- image-classification
- image-captioning

---

# Poster2Plot

An image captioning model to generate movie/t.v show plot from poster. It generates decent plots but is no way perfect. We are still working on improving the model.

## Live demo on Hugging Face Spaces: https://huggingface.co/spaces/deepklarity/poster2plot

# Model Details

The base model uses a Vision Transformer (ViT) model as an image encoder and GPT-2 as a decoder.

We used the following models:

* Encoder: [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k)
* Decoder: [gpt2](https://huggingface.co/gpt2)

# Datasets

Publicly available IMDb datasets were used to train the model.

# How to use

## In PyTorch

```python
import torch
import re
import requests
from PIL import Image
from transformers import AutoTokenizer, AutoFeatureExtractor, VisionEncoderDecoderModel

# Pattern to ignore all the text after 2 or more full stops
regex_pattern = "[.]{2,}"


def post_process(text):
    try:
        text = text.strip()
        text = re.split(regex_pattern, text)[0]
    except Exception as e:
        print(e)
        pass
    return text


def predict(image, max_length=64, num_beams=4):
    pixel_values = feature_extractor(images=image, return_tensors="pt").pixel_values
    pixel_values = pixel_values.to(device)

    with torch.no_grad():
        output_ids = model.generate(
            pixel_values,
            max_length=max_length,
            num_beams=num_beams,
            return_dict_in_generate=True,
        ).sequences

    preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
    pred = post_process(preds[0])

    return pred


model_name_or_path = "deepklarity/poster2plot"
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

# Load model.

model = VisionEncoderDecoderModel.from_pretrained(model_name_or_path)
model.to(device)
print("Loaded model")

feature_extractor = AutoFeatureExtractor.from_pretrained(model.encoder.name_or_path)
print("Loaded feature_extractor")

tokenizer = AutoTokenizer.from_pretrained(model.decoder.name_or_path, use_fast=True)
if model.decoder.name_or_path == "gpt2":
    tokenizer.pad_token = tokenizer.eos_token

print("Loaded tokenizer")

url = "https://upload.wikimedia.org/wikipedia/en/2/26/Moana_Teaser_Poster.jpg"
with Image.open(requests.get(url, stream=True).raw) as image:
    pred = predict(image)

print(pred)

```