Spaces:
Running
Running
Rishi Desai
commited on
Commit
·
524c601
1
Parent(s):
aa6982c
readme + trigger word
Browse files- README.md +20 -40
- caption.py +1 -1
README.md
CHANGED
@@ -1,18 +1,18 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
-
|
15 |
-
-
|
16 |
|
17 |
## Installation
|
18 |
|
@@ -20,7 +20,6 @@ AutoCaptioner creates detailed, principled image captions for your LoRA dataset.
|
|
20 |
- Python 3.11 or higher
|
21 |
- [Together API](https://together.ai/) account and API key
|
22 |
|
23 |
-
|
24 |
### Setup
|
25 |
|
26 |
1. Create the virtual environment:
|
@@ -30,9 +29,7 @@ AutoCaptioner creates detailed, principled image captions for your LoRA dataset.
|
|
30 |
python -m pip install -r requirements.txt
|
31 |
```
|
32 |
|
33 |
-
2.
|
34 |
-
|
35 |
-
3. Run inference on one set of images:
|
36 |
|
37 |
```bash
|
38 |
python main.py --input examples/ --output output/
|
@@ -43,8 +40,7 @@ AutoCaptioner creates detailed, principled image captions for your LoRA dataset.
|
|
43 |
|
44 |
- `--input` (str): Directory containing images to caption.
|
45 |
- `--output` (str): Directory to save images and captions (defaults to input directory).
|
46 |
-
- `--
|
47 |
-
- `--batch_images` (flag): Process images in batches by category.
|
48 |
</details>
|
49 |
|
50 |
|
@@ -55,31 +51,15 @@ Launch a user-friendly web interface for captioning and prompt optimization:
|
|
55 |
python demo.py
|
56 |
```
|
57 |
|
58 |
-
###
|
59 |
-
|
60 |
-
- High-accuracy image captioning with detailed contextual descriptions
|
61 |
-
- Consistent character descriptions when using the outfit flag
|
62 |
-
- Batch processing for large image collections
|
63 |
-
- Optimized for AI model training datasets
|
64 |
-
- Web interface for easy use
|
65 |
-
|
66 |
-
## How It Works
|
67 |
-
|
68 |
-
AutoCaptioner leverages the Llama-4-Maverick model through the Together AI platform to:
|
69 |
-
1. Analyze the visual content of your images
|
70 |
-
2. Generate detailed, structured captions
|
71 |
-
3. Save the captions as text files alongside your images
|
72 |
-
|
73 |
-
## Notes
|
74 |
- Images are processed individually in standard mode
|
75 |
- For large collections, batch processing by category is recommended
|
76 |
- Each caption is saved as a .txt file with the same name as the image
|
77 |
|
78 |
### Troubleshooting
|
79 |
|
80 |
-
- **API errors**: Ensure your Together API key is set
|
81 |
-
- **
|
82 |
-
- **Memory issues**: For very large images, try processing in smaller batches
|
83 |
|
84 |
### Examples
|
85 |
|
|
|
1 |
+
---
|
2 |
+
title: LoRACaptioner
|
3 |
+
emoji: 🤠
|
4 |
+
colorFrom: red
|
5 |
+
colorTo: green
|
6 |
+
sdk: gradio
|
7 |
+
sdk_version: 5.25.2
|
8 |
+
app_file: demo.py
|
9 |
+
pinned: false
|
10 |
+
---
|
11 |
+
|
12 |
+
# LoRACaptioner
|
13 |
+
|
14 |
+
- **Image Captioning**: Automatically generate detailed and structured captions for your LoRA dataset.
|
15 |
+
- **Prompt Optimization**: Enhance prompts during inference to achieve high-quality outputs.
|
16 |
|
17 |
## Installation
|
18 |
|
|
|
20 |
- Python 3.11 or higher
|
21 |
- [Together API](https://together.ai/) account and API key
|
22 |
|
|
|
23 |
### Setup
|
24 |
|
25 |
1. Create the virtual environment:
|
|
|
29 |
python -m pip install -r requirements.txt
|
30 |
```
|
31 |
|
32 |
+
2. Run inference on one set of images:
|
|
|
|
|
33 |
|
34 |
```bash
|
35 |
python main.py --input examples/ --output output/
|
|
|
40 |
|
41 |
- `--input` (str): Directory containing images to caption.
|
42 |
- `--output` (str): Directory to save images and captions (defaults to input directory).
|
43 |
+
- `--batch_images` (flag): Caption images in batches by category.
|
|
|
44 |
</details>
|
45 |
|
46 |
|
|
|
51 |
python demo.py
|
52 |
```
|
53 |
|
54 |
+
### Notes
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
55 |
- Images are processed individually in standard mode
|
56 |
- For large collections, batch processing by category is recommended
|
57 |
- Each caption is saved as a .txt file with the same name as the image
|
58 |
|
59 |
### Troubleshooting
|
60 |
|
61 |
+
- **API errors**: Ensure your Together API key is set and has funds
|
62 |
+
- **Image formats**: Only .png, .jpg, .jpeg, and .webp files are supported
|
|
|
63 |
|
64 |
### Examples
|
65 |
|
caption.py
CHANGED
@@ -4,7 +4,7 @@ import os
|
|
4 |
from together import Together
|
5 |
|
6 |
MODEL_ID = "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8"
|
7 |
-
TRIGGER_WORD = "
|
8 |
|
9 |
def get_system_prompt():
|
10 |
return f"""Automated Image Captioning (for LoRA Training)
|
|
|
4 |
from together import Together
|
5 |
|
6 |
MODEL_ID = "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8"
|
7 |
+
TRIGGER_WORD = "tr1gg3r"
|
8 |
|
9 |
def get_system_prompt():
|
10 |
return f"""Automated Image Captioning (for LoRA Training)
|