Spaces:
Running
Running
File size: 2,631 Bytes
4818cbb 7a920b1 4818cbb 7a920b1 4818cbb 7a920b1 |
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 |
import base64
from io import BytesIO
import pandas as pd
import ast
def pil_to_base64(pil_image):
"""Convert PIL Image to base64 string"""
buffered = BytesIO()
pil_image.save(buffered, format="PNG")
img_str = base64.b64encode(buffered.getvalue()).decode()
return f"data:image/png;base64,{img_str}"
def image_to_base64(img_bytes):
"""Convert image bytes to base64 string with MIME type
Handles both:
- Dict format: {'bytes': b'...', 'path': None}
- String format from CSV: "{'bytes': b'...', 'path': None}"
- Raw bytes
"""
if isinstance(img_bytes, str):
img_dict = ast.literal_eval(img_bytes)
if isinstance(img_dict, dict) and "bytes" in img_dict:
img_bytes = img_dict["bytes"]
elif isinstance(img_bytes, dict) and "bytes" in img_bytes:
img_bytes = img_bytes["bytes"]
# Encode to base64
b64_string = base64.b64encode(img_bytes).decode("utf-8")
return f"data:image/jpeg;base64,{b64_string}"
def create_training_example(row):
"""Create a training example with both classification and description tasks"""
img_b64 = image_to_base64(row["image"])
user_prompt = "Analyze this fashion product image and provide: 1) Master category, 2) Gender, 3) Sub-category, and 4) A detailed description."
assistant_response = f"""
Master Category: {row['masterCategory']}
Gender: {row['gender']}
Sub-category: {row['subCategory']}
Description: This is a {row['gender'].lower()} {row['subCategory'].lower()} from the {row['masterCategory'].lower()} category."""
return {
"messages": [
{
"role": "system",
"content": "You are a fashion product analyst. Classify products and generate detailed descriptions based on images.",
},
{
"role": "user",
"content": [
{"type": "image_url", "image_url": {"url": img_b64}},
{"type": "text", "text": user_prompt},
],
},
{"role": "assistant", "content": assistant_response},
]
}
def load_test_data(csv_path="../data/test.csv"):
"""
Load test data for inference from CSV file
Args:
csv_path: Path to the test CSV file (default: 'data/test.csv')
Returns:
pd.DataFrame: Test data with columns for inference
"""
# Load test data from CSV
df_test = pd.read_csv(csv_path)
print(f"Loaded {len(df_test)} test examples from {csv_path}")
print(f"Columns: {df_test.columns.tolist()}")
return df_test
|