<h3 id="crotch-detector-(pussy-+-pubes-+-anus)-65ldc98fr"><strong>Crotch Detector (<span style="color:rgb(230, 73, 128)">Pussy</span> + Pubes + <span style="color:rgb(181, 102, 45)">Anus</span>)</strong></h3><p><em>EDIT (2025-06-25): scroll down for sample ComfyUI workflow (image contains workflow, save it and drag/drop it onto your ComfyUI window to import)</em></p><p>I was unhappy with the detection performance of the available adetailers and wanted something better suiting my workflow. So I used the masks I have generated during the last year or so to train this model starting from yolo11l.</p><p>As I am happy with the detection performance I would like to share this here so others can use it, too.</p><ul><li><p>Architecture: <strong>YOLO11L</strong></p></li><li><p>Imgsz: <strong>832</strong></p></li><li><p>Style: <strong>SEGM</strong></p></li><li><p><strong>Training Dataset: 1,331 semi-realistic images</strong></p></li></ul><h3 id="what-will-it-detect-takcr8stt"><strong>What will it detect?</strong></h3><p>It detects the female intimate area including the pubic hair and anus in semi-realistic images. I believe it could work in realistic and anime images, too, but I haven't tried.</p><h3 id="what-is-this-for-nuupo9sp8"><strong>What is this for?</strong></h3><p>You can use that for automating the inpainting process of the intimate area or to generate masks of pussies.</p><h3 id="how-can-this-be-used-ibe11mtpa"><strong>How can this be used?</strong></h3><ul><li><p>With python and the <a target="_blank" rel="ugc" href="https://docs.ultralytics.com/modes/predict/">ultralytics module</a> (scroll down to the python code)</p></li><li><p>With <a target="_blank" rel="ugc" href="https://github.com/comfyanonymous/ComfyUI">ComfyUI</a> and the <a target="_blank" rel="ugc" href="https://github.com/ltdrdata/ComfyUI-Impact-Pack">Impact Pack</a> + <a target="_blank" rel="ugc" href="https://github.com/ltdrdata/ComfyUI-Impact-Subpack">Impact Subpack</a></p></li><li><p>Probably in <a target="_blank" rel="ugc" href="https://github.com/AUTOMATIC1111/stable-diffusion-webui">A1111</a> but I have not tested that</p></li></ul><p></p><h3 id="validation-results-yr5hf0vca"><strong>Validation Results</strong></h3><h3 id="-mcd2jlxj0"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/63d98645-69d0-445e-a0d1-b0c5f89d5b2b/width=525/63d98645-69d0-445e-a0d1-b0c5f89d5b2b.jpeg" /></h3><h3 id="training-progress-d9xhg3ylw"><strong>Training Progress</strong></h3><h3 id="-pdj5tsd3b"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f9207cf5-815c-4f83-97a2-5939319007e8/width=525/f9207cf5-815c-4f83-97a2-5939319007e8.jpeg" /></h3><h3 id="sample-comfyui-workflow-feqwzv7ef"><strong>Sample ComfyUI Workflow</strong></h3><h3 id="thanks-and-acknowledgements-m1kcfi8er"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/d1b8bf33-003c-476f-9535-3cb0b6176092/width=525/d1b8bf33-003c-476f-9535-3cb0b6176092.jpeg" /><strong>Thanks and Acknowledgements</strong></h3><p>I would like to thank Ultralytics for providing the base model weights and their training scripts. And I would like to thank Team Comfy for providing an excellent UI for image generation and last but not least CivitAI for hosting all these fine models and resources.</p><h3 id="final-note-5rhxoc6io"><strong>Final Note</strong></h3><p>In case you have questions leave a comment here or contact <a target="_blank" rel="ugc" href="https://www.deviantart.com/0xuberlord">me on Deviant Art</a> (I am not here that often so I might not respond immediately).</p>

Files with SHA256: 8ca79171ac17d469213c20c328fdbd05395288c954248ef018b7fd1ce0df23ad