mohammad-shirkhani
commited on
Commit
•
ccd4a42
1
Parent(s):
2b0afb0
Update README.md
Browse files
README.md
CHANGED
@@ -2,28 +2,97 @@
|
|
2 |
|
3 |
## Model Overview
|
4 |
|
5 |
-
This model pipeline is designed to generate images from Persian text descriptions by translating the Persian text into English and then using a fine-tuned Stable Diffusion model to generate the corresponding image. The pipeline combines two models: a translation model (`mohammad-shirkhani/finetune_persian_to_english_mt5_base_summarize_on_celeba_hq`) and an image generation model (`ebrahim-k/Stable-Diffusion-1_5-FT-celeba_HQ_en`).
|
6 |
|
7 |
## Model Details
|
8 |
|
9 |
### Translation Model
|
10 |
- **Model Name**: `mohammad-shirkhani/finetune_persian_to_english_mt5_base_summarize_on_celeba_hq`
|
11 |
- **Architecture**: mT5
|
12 |
-
- **Purpose**: This model
|
13 |
|
14 |
### Image Generation Model
|
15 |
- **Model Name**: `ebrahim-k/Stable-Diffusion-1_5-FT-celeba_HQ_en`
|
16 |
- **Architecture**: Stable Diffusion 1.5
|
17 |
-
- **Purpose**: This model generates high-quality images from
|
18 |
|
19 |
## Pipeline Description
|
20 |
|
21 |
-
The pipeline
|
22 |
|
23 |
1. **Text Translation**: The Persian input text is translated into English using the mT5-based translation model.
|
24 |
-
2. **Image Generation**: The translated English text is then
|
25 |
|
26 |
-
###
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
|
28 |
```python
|
29 |
from IPython.display import display
|
@@ -39,3 +108,4 @@ display(image)
|
|
39 |
persian_text2 = "این مرد جذاب دارای موهای قهوه ای ، سوزش های جانبی ، دهان کمی باز و کیسه های زیر چشم است.این فرد جذاب دارای کیسه های زیر چشم ، سوزش های جانبی و دهان کمی باز است."
|
40 |
image2 = persian_to_image_model(persian_text2)
|
41 |
display(image2)
|
|
|
|
2 |
|
3 |
## Model Overview
|
4 |
|
5 |
+
This model pipeline is designed to generate images from Persian text descriptions. It works by first translating the Persian text into English and then using a fine-tuned Stable Diffusion model to generate the corresponding image. The pipeline combines two models: a translation model (`mohammad-shirkhani/finetune_persian_to_english_mt5_base_summarize_on_celeba_hq`) and an image generation model (`ebrahim-k/Stable-Diffusion-1_5-FT-celeba_HQ_en`).
|
6 |
|
7 |
## Model Details
|
8 |
|
9 |
### Translation Model
|
10 |
- **Model Name**: `mohammad-shirkhani/finetune_persian_to_english_mt5_base_summarize_on_celeba_hq`
|
11 |
- **Architecture**: mT5
|
12 |
+
- **Purpose**: This model translates Persian text into English. It has been fine-tuned on the CelebA-HQ dataset for summarization tasks, making it effective for translating descriptions of facial features.
|
13 |
|
14 |
### Image Generation Model
|
15 |
- **Model Name**: `ebrahim-k/Stable-Diffusion-1_5-FT-celeba_HQ_en`
|
16 |
- **Architecture**: Stable Diffusion 1.5
|
17 |
+
- **Purpose**: This model generates high-quality images from English text produced by the translation model. It has been fine-tuned on the CelebA-HQ dataset, which makes it particularly effective for generating realistic human faces based on text descriptions.
|
18 |
|
19 |
## Pipeline Description
|
20 |
|
21 |
+
The pipeline operates through the following steps:
|
22 |
|
23 |
1. **Text Translation**: The Persian input text is translated into English using the mT5-based translation model.
|
24 |
+
2. **Image Generation**: The translated English text is then used to generate the corresponding image with the Stable Diffusion model.
|
25 |
|
26 |
+
### Code Implementation
|
27 |
+
|
28 |
+
#### 1. Install Required Libraries
|
29 |
+
|
30 |
+
```python
|
31 |
+
!pip install transformers diffusers accelerate torch
|
32 |
+
```
|
33 |
+
#### 2. Import Necessary Libraries
|
34 |
+
|
35 |
+
```python
|
36 |
+
import torch
|
37 |
+
from transformers import MT5ForConditionalGeneration, T5Tokenizer
|
38 |
+
from diffusers import StableDiffusionPipeline
|
39 |
+
```
|
40 |
+
|
41 |
+
#### 3. Set Device (GPU or CPU)
|
42 |
+
This code determines whether the pipeline should use a GPU (if available) or fallback to a CPU.
|
43 |
+
|
44 |
+
```python
|
45 |
+
# Determine the device: GPU if available, otherwise CPU
|
46 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
47 |
+
print(f"Using device: {device}")
|
48 |
+
```
|
49 |
+
|
50 |
+
#### 4. Define and Load the Persian-to-Image Model Class
|
51 |
+
The following class handles both translation and image generation tasks.
|
52 |
+
|
53 |
+
```python
|
54 |
+
# Define the model class
|
55 |
+
class PersianToImageModel:
|
56 |
+
def __init__(self, translation_model_name, image_model_name, device):
|
57 |
+
self.device = device
|
58 |
+
|
59 |
+
# Load translation model
|
60 |
+
self.translation_model = MT5ForConditionalGeneration.from_pretrained(translation_model_name).to(device)
|
61 |
+
self.translation_tokenizer = T5Tokenizer.from_pretrained(translation_model_name)
|
62 |
+
|
63 |
+
# Load image generation model
|
64 |
+
self.image_model = StableDiffusionPipeline.from_pretrained(image_model_name).to(device)
|
65 |
+
|
66 |
+
def translate_text(self, persian_text):
|
67 |
+
input_ids = self.translation_tokenizer.encode(persian_text, return_tensors="pt").to(self.device)
|
68 |
+
translated_ids = self.translation_model.generate(input_ids, max_length=512, num_beams=4, early_stopping=True)
|
69 |
+
translated_text = self.translation_tokenizer.decode(translated_ids[0], skip_special_tokens=True)
|
70 |
+
return translated_text
|
71 |
+
|
72 |
+
def generate_image(self, english_text):
|
73 |
+
image = self.image_model(english_text).images[0]
|
74 |
+
return image
|
75 |
+
|
76 |
+
def __call__(self, persian_text):
|
77 |
+
# Translate Persian text to English
|
78 |
+
english_text = self.translate_text(persian_text)
|
79 |
+
print(f"Translated Text: {english_text}")
|
80 |
+
|
81 |
+
# Generate and return image
|
82 |
+
return self.generate_image(english_text)
|
83 |
+
```
|
84 |
+
#### 5. Instantiate the Model
|
85 |
+
The following code snippet demonstrates how to instantiate the combined model.
|
86 |
+
|
87 |
+
```python
|
88 |
+
# Instantiate the combined model
|
89 |
+
translation_model_name = 'mohammad-shirkhani/finetune_persian_to_english_mt5_base_summarize_on_celeba_hq'
|
90 |
+
image_model_name = 'ebrahim-k/Stable-Diffusion-1_5-FT-celeba_HQ_en'
|
91 |
+
|
92 |
+
persian_to_image_model = PersianToImageModel(translation_model_name, image_model_name, device)
|
93 |
+
```
|
94 |
+
#### 6. Example Usage of the Model
|
95 |
+
Below are examples of how to use the model to generate images from Persian text.
|
96 |
|
97 |
```python
|
98 |
from IPython.display import display
|
|
|
108 |
persian_text2 = "این مرد جذاب دارای موهای قهوه ای ، سوزش های جانبی ، دهان کمی باز و کیسه های زیر چشم است.این فرد جذاب دارای کیسه های زیر چشم ، سوزش های جانبی و دهان کمی باز است."
|
109 |
image2 = persian_to_image_model(persian_text2)
|
110 |
display(image2)
|
111 |
+
```
|