how to make tool like remove.bg

@salman-qureshi Check the courses, there is a full course about APIs

@salman-qureshi He did already.

@ssadvisor i have course where he made removing bg tool

@salman-qureshi You mean the blog post. And, if you want to turn it into a tool, go over the steps @admin mentioned before in this thread.

Plus, you can try these courses; they will help you with every step.

@admin i am trying to an api but it get error every time please help me

import os
import io
import uuid
from fastapi import FastAPI, File, UploadFile, HTTPException
from PIL import Image
import numpy as np
from skimage import transform
import torch
from torch.autograd import Variable
from model import U2NET
from pydantic import BaseModel

app = FastAPI()

# Load the trained model
model_name = 'u2net'
model_dir = os.path.join(os.path.dirname(__file__), 'saved_models', model_name, f'{model_name}.pth')
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
net = U2NET(3, 1)
net.load_state_dict(torch.load(model_dir, map_location=device))
net.eval()
net.to(device)

# Function to save output image
def save_output(image_name, output_name, pred, output_dir, suffix='imagebgremoval.com'):
    predict_np = pred.cpu().squeeze().numpy() * 255
    im = Image.fromarray(predict_np.astype(np.uint8)).convert('RGB')
    image = Image.open(image_name)
    imo = im.resize((image.width, image.height))
    imo.save(os.path.join(output_dir, f"{os.path.splitext(output_name)[0]}_{suffix}.png"))

# Function to remove background
def remove_background(image):
    image = transform.resize(image, (320, 320), mode='constant')
    image = (image - [0.485, 0.456, 0.406]) / [0.229, 0.224, 0.225]
    image = image.transpose((2, 0, 1))
    image = np.expand_dims(image, 0)
    image = torch.tensor(image, dtype=torch.float32).to(device)
    with torch.no_grad():
        d1, _, _, _, _, _, _ = net(image)
        pred = torch.sigmoid(d1[:, 0, :, :])
    return pred

# Define response model
class BackgroundRemovalResponse(BaseModel):
    message: str

# API endpoint for background removal
@app.post("/remove_bg/")
async def process_image(image_file: UploadFile = File(...)):
    try:
        # Validate file type
        if not image_file.content_type.startswith('image'):
            raise HTTPException(status_code=400, detail="File is not an image.")

        # Read and preprocess the image
        image = Image.open(io.BytesIO(await image_file.read())).convert('RGB')
        if image.mode != "RGB":
            raise HTTPException(status_code=400, detail="Image mode is not RGB.")
        
        # Remove background
        pred = remove_background(np.array(image))

        # Save output image
        output_dir = os.path.join(os.path.dirname(__file__), 'static', 'results')
        unique_filename = str(uuid.uuid4())
        save_output(image, unique_filename + '.png', pred, output_dir)

        return BackgroundRemovalResponse(message=f"Background removed successfully. Output file: {unique_filename}_imagebgremoval.com.png")
    except Exception as e:
        raise HTTPException(status_code=500, detail="Failed to remove background.") from e

@salman-qureshi whats the error you're getting?

@husein @admin sometime internal error some time bg not removing from image some time whole image become blueish that's all 

this result i am getting but script working perfectly but api is not 

@salman-qureshi you mean when you turn into an API you get different results?

@admin yes

@admin you said in your video that you will also give api code also and you do not give kindly help me

@salman-qureshi OpenAI will never give you the same result for the same prompt.

@ssadvisor @admin can you give me right prompt please

@salman-qureshi There is no right prompt. There are good prompts, but also these good prompts sometime give different results on each run.

@husein @SSAdvisor @admin so how can i turn these into api please tell me

@salman-qureshi Friend, if you want to build an API, you should at least learn the basics of Python APIs and how they work. Do you know python? did you built an API before?

@salman-qureshi you need to review the courses. This is explained there.

@admin but you said in your video that you will give api also but you not give

brother

@salman-qureshi link to the video you reference?

@ssadvisor @admin https://youtu.be/HNu6FgWGcYM?si=PsSUhqZNhfJaqNvj

@salman-qureshi he gave you the code. https://learnwithhasan.com/remove-image-background-with-python/