Unofficial mirror of the Qwen Image 2.1 Object Remover BBox Turbo LoRA by prithivMLmods.<h2 id="what-it-does">What it does</h2>Turbo-optimized LoRA adapter for Qwen-Image-2.1, designed to remove unwanted objects inside user-defined bounding boxes while preserving surrounding textures, lighting, shadows, perspective, and overall image consistency. Supports Turbo and standard workflows.<h2 id="model-details">Model details</h2><ul><li>Base model: <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image-2.1">Qwen/Qwen-Image-2.1</a></li><li>Adapter: Qwen-Image-2.1-Object-Remover-Bbox-turbo</li><li>Type: LoRA / Adapter</li><li>Status: Experimental</li><li>Inference: Turbo / Standard</li><li>Creator: <a target="_blank" rel="ugc" href="https://huggingface.co/prithivMLmods">prithivMLmods</a></li></ul><h2 id="training-specifications">Training specifications</h2><ul><li>Dataset: 80 pairs of high-quality images with bounding box annotations and manually manipulated resultant images</li><li>Save precision: BF16</li><li>Learning rate: 1e-4</li><li>Optimizer: AdamW</li><li>Network dimension / rank: 16</li><li>Total steps: 4000</li><li>Trigger prompt: <code>Remove the red highlighted object from the scene</code></li></ul><h2 id="usage-and-inference">Usage and inference</h2>Load the LoRA adapter with Qwen-Image-2.1 and provide an image with the target object indicated by a clear bounding box.Trigger prompt: <code>Remove the red highlighted object from the scene</code>This adapter is optimized for fast Turbo inference while remaining compatible with standard workflows. For Turbo inference, follow the recommended Qwen-Image-2.1 Turbo configuration.<h3 id="diffusers-example">Diffusers example</h3><pre><code>pip install -U diffusers transformers accelerate

import torch
from diffusers import DiffusionPipeline
from diffusers.utils import load_image

pipe = DiffusionPipeline.from_pretrained(
 "Qwen/Qwen-Image-2.1",
 dtype=torch.bfloat16,
 device_map="cuda",
)
pipe.load_lora_weights(
 "prithivMLmods/Qwen-Image-2.1-Object-Remover-Bbox-turbo"
)
prompt = "Remove the red highlighted object from the scene"
image = load_image("input.png")
result = pipe(image=image, prompt=prompt).images[0]</code></pre><h2 id="comparison-summary">Comparison summary</h2><pre><code>Scenario Base model result LoRA result / observation Steps
Shadows May struggle to retain shadows Better preservation of surrounding shadows 40
Cats One cat may remain unremoved Better handling of the marked bounding boxes 40
Objects May remove unmarked objects; struggles Experimental behavior with more than two objects 40
Multiple BBox May struggle with more than one box Designed for bounding-box object removal 40</code></pre><h2 id="visual-comparisons-from-the-original-readme">Visual comparisons from the original README</h2><h3 id="1.-shadows">1. Shadows</h3>Input<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/XyF1CJV0yAlOchE53LW8_.jpeg" alt="Input - Shadows" width="320" />Base model<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/-zu8x11Wv-9wA1JBz3You.png" alt="Base model - Shadows" width="320" />With LoRA<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/CPE76r-vUksqeZLdSWMC_.png" alt="LoRA - Shadows" width="320" />Base model without LoRA can struggle to retain shadows after removal. Total steps: 40.<h3 id="2.-cats">2. Cats</h3>Input<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/dHKtDENkrtKenQnVDK5-F.png" alt="Input - Cats" width="320" />Base model<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/wLYk70jFNzYSYwZIL8Dq8.png" alt="Base model - Cats" width="320" />With LoRA<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/3Paj-1Bbos_0Pcp97o1yu.png" alt="LoRA - Cats" width="320" />Base model without LoRA can leave one cat unremoved in multi-box cases. Total steps: 40.<h3 id="3.-objects">3. Objects</h3>Input<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/UtGcANw9ssjqmrlFyzW4u.png" alt="Input - Objects" width="320" />Base model<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/FER98fUr-lcvUZrrY1A6G.png" alt="Base model - Objects" width="320" />With LoRA<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/hEwBd3LKNcDP9cnmosp7z.png" alt="LoRA - Objects" width="320" />The original comparison notes that unmarked objects may be removed and that more than two objects can be difficult. Total steps: 40.<h3 id="4.-multiple-bounding-boxes">4. Multiple bounding boxes</h3>Input<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/p1Tdm4b04_VDAsokJ9COy.png" alt="Input - Multiple Bbox" width="320" />Base model<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/63sNmdizhxE9VUhqWAfry.png" alt="Base model - Multiple Bbox" width="320" />With LoRA<img src="https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/hW_uWzMOL8MUA0dQUq_Oy.png" alt="LoRA - Multiple Bbox" width="320" />Base model without LoRA can struggle to remove more than one object in a bounding-box scenario. Total steps: 40.<h2 id="limitations">Limitations</h2>Experimental release. Results may vary with complex backgrounds, large or overlapping objects, fine structures and textures, reflections and transparent objects, difficult lighting or perspective, and ambiguous or poorly positioned bounding boxes.<h2 id="original-source-and-license">Original source and license</h2>Original repository: <a target="_blank" rel="ugc" href="https://huggingface.co/prithivMLmods/Qwen-Image-2.1-Object-Remover-Bbox-turbo">prithivMLmods/Qwen-Image-2.1-Object-Remover-Bbox-turbo</a>License: Qwen Research License Agreement — non-commercial use only. This is an unofficial mirror and is not created or endorsed by the original author. Preserve the original license and attribution notices. Do not sell this model or merges made from it. Built with Qwen.

Files with SHA256: 7035eba6f25cdbd0c780027372710c9bebebdd814baaa79d48ce9d5ca80d6449