Flask API本地运行正常但Heroku部署后出现400请求错误：大Base64图片字符串传输问题

Hey there! Let's fix that 400 error you're hitting on Heroku with your Flask API. You already nailed the root cause—shoving a massive Base64 image string into request headers is a no-go for production servers like Heroku's, even though your local setup lets it slide. Here are the most practical solutions to get your image data where it needs to go:

Solution 1: Send Base64 data in the POST request body (instead of headers) #

Most web servers (including Heroku's) enforce strict limits on header sizes (often just a few KB) to prevent abuse, but they're much more lenient with request bodies. You can package your Base64 string and unique ID into a JSON payload and send it in the body of a POST request.

Flask API Code Example #

from flask import Flask, request, jsonify
import base64
import io
from PIL import Image

app = Flask(__name__)
# Optional: Set a max request body size (adjust based on your image needs)
app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024  # 16MB limit

@app.route('/detect-food', methods=['POST'])
def detect_food():
    try:
        # Parse JSON payload from request body
        payload = request.get_json()
        img_b64 = payload.get('image')
        unique_id = payload.get('unique_id')

        # Validate required fields
        if not img_b64 or not unique_id:
            return jsonify({"error": "Missing image or unique ID"}), 400

        # Decode Base64 and process the image
        img_data = base64.b64decode(img_b64)
        img = Image.open(io.BytesIO(img_data))

        # Run your TensorFlow Lite detection logic here
        # ...
        # Save results to your database
        # ...

        return jsonify({"status": "success", "unique_id": unique_id})
    except Exception as e:
        return jsonify({"error": str(e)}), 400

if __name__ == '__main__':
    app.run()

Client Request Example (Python) #

import requests
import base64

# Encode local image to Base64
with open('your-food-image.jpg', 'rb') as img_file:
    img_b64 = base64.b64encode(img_file.read()).decode('utf-8')

# Send payload in request body
payload = {
    "image": img_b64,
    "unique_id": "your-business-related-id"
}

response = requests.post('https://your-heroku-app.herokuapp.com/detect-food', json=payload)
print(response.json())

Solution 2: Upload the image directly as a file (more efficient) #

Base64 adds ~30% extra size to your image, which is unnecessary. Instead, use multipart/form-data to upload the raw image file along with your unique ID. This is the preferred method for large media files.

Flask API Code Example #

from flask import Flask, request, jsonify
from PIL import Image

app = Flask(__name__)
app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024  # 16MB limit

@app.route('/detect-food', methods=['POST'])
def detect_food():
    try:
        # Get image file and unique ID from form data
        if 'image' not in request.files:
            return jsonify({"error": "No image file provided"}), 400
        
        img_file = request.files['image']
        unique_id = request.form.get('unique_id')

        if not unique_id:
            return jsonify({"error": "Missing unique ID"}), 400

        # Process the image directly from the file stream
        img = Image.open(img_file.stream)

        # Run TensorFlow Lite detection logic
        # ...
        # Save results to database
        # ...

        return jsonify({"status": "success", "unique_id": unique_id})
    except Exception as e:
        return jsonify({"error": str(e)}), 400

if __name__ == '__main__':
    app.run()

Client Request Example (Python) #

import requests

# Prepare form data and file
payload = {
    "unique_id": "your-business-related-id"
}
files = {
    "image": open('your-food-image.jpg', 'rb')
}

response = requests.post('https://your-heroku-app.herokuapp.com/detect-food', data=payload, files=files)
print(response.json())

Why this works locally but not on Heroku #

Your local Flask server (Werkzeug) has very loose limits on request headers for development convenience. But Heroku uses production-grade servers like Gunicorn, which have strict default header size limits (usually around 8KB) to protect against attacks. Your Base64 image string was way over that limit, triggering the 400 error.

Bonus Tip for Extra Large Images #

If your images exceed Heroku's default request body limit (around 100MB), you can:

1. Upload the image to a cloud storage service first (like AWS S3 or Google Cloud Storage)
2. Send the image's URL and unique ID to your Flask API
3. Have the API fetch the image from the URL for processing

内容的提问来源于stack exchange，提问作者Siddharth Agrawal