{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-29T09:40:48.255Z","headline":"FeyNoBg 发布：开源自动背景去除模型，在四项基准上达到 SOTA","description":"Feyn Labs 推出 FeyNoBg，一个用于自动背景去除的 SOTA 模型。它在八个基准测试中的四项上取得最佳 S-measure 分数，其余四项与领先者差距在 2% 以内。该模型基于 BiRefNet 架构，参数量从 222M 扩展至 263M，同时开源了训练库 NoBg，模型和代码分别可在 Hugging Face 和 GitHub 获取。","url":"https://www.aioga.com/news/cms47q4jj00mxroep0x4at5sf/","mainEntityOfPage":"https://www.aioga.com/news/cms47q4jj00mxroep0x4at5sf/","datePublished":"2026-07-28T04:57:13.190Z","dateModified":"2026-07-28T04:57:13.190Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://usefeyn.com/blog/feynobg","https://aihot.virxact.com/items/cms47q4jj00mxroep0x4at5sf"],"canonicalUrl":"https://www.aioga.com/news/cms47q4jj00mxroep0x4at5sf/","directAnswer":{"@type":"Answer","text":"Feyn Labs 发布开源背景去除模型 FeyNoBg，并同步开源训练库 NoBg。材料称，该模型在八项基准中的四项取得最佳 S-measure，其余四项与领先者差距不超过 2%。","url":"https://www.aioga.com/news/cms47q4jj00mxroep0x4at5sf/","dateCreated":"2026-07-28T04:57:13.190Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"usefeyn.com source article","url":"https://usefeyn.com/blog/feynobg","datePublished":"2026-07-28T04:57:13.190Z","provider":{"@type":"Organization","name":"usefeyn.com","url":"https://usefeyn.com/blog/feynobg"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cms47q4jj00mxroep0x4at5sf","datePublished":"2026-07-28T04:57:13.190Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cms47q4jj00mxroep0x4at5sf"}}],"aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","originalPublisher":{"name":"usefeyn.com","url":"https://usefeyn.com/blog/feynobg"},"article":{"id":"cms47q4jj00mxroep0x4at5sf","slug":"cms47q4jj00mxroep0x4at5sf","url":"https://www.aioga.com/news/cms47q4jj00mxroep0x4at5sf/","title":"FeyNoBg 发布：开源自动背景去除模型，在四项基准上达到 SOTA","title_en":"Show HN： FeyNoBg - 自动背景去除模型及训练库","summary":"Feyn Labs 推出 FeyNoBg，一个用于自动背景去除的 SOTA 模型。它在八个基准测试中的四项上取得最佳 S-measure 分数，其余四项与领先者差距在 2% 以内。该模型基于 BiRefNet 架构，参数量从 222M 扩展至 263M，同时开源了训练库 NoBg，模型和代码分别可在 Hugging Face 和 GitHub 获取。","source":"Hacker News 热门（buzzing.cc 中文翻译）","sourceUrl":"https://usefeyn.com/blog/feynobg","aiHotUrl":"https://aihot.virxact.com/items/cms47q4jj00mxroep0x4at5sf","publishedAt":"2026-07-28T04:57:13.190Z","category":"模型更新","score":71,"selected":true,"articleBody":["We’re introducing FeyNoBg, a state-of-the-art model for automatic background removal. Across eight benchmarks, it posts the best published S-measure on four and comes within 2% of the leader on the rest.","Tests salient-object masks in ultra-high-resolution images, including 4K and 8K scenes.","We’re also releasing NoBg, the library we used to train our model. Use NoBg to run FeyNoBg or train your own background removal model.","Both are open source. Download FeyNoBg on Hugging Face：https://huggingface.co/feyninc/FeyNobg, or build with NoBg on GitHub：https://github.com/feyninc/nobg.","In your computer, images are stored as grids of pixels. The goal of a background removal algorithm is to predict an opacity value for each pixel such that background pixels become transparent, foreground pixels remain opaque, and boundary pixels become translucent.","Producing this opacity map requires two skills. First, the model has to separate the foreground from the background. When the subject stands against a plain backdrop, this can be done easily by comparing colors. However, in low contrast, crowded, or camouflaged images, the model has to use shape, texture, and context to recognize which pixels form the subject.","Second, the model has to trace the foreground’s boundary. In simple images, a sharp change in color or texture provides a strong edge signal. But real boundaries are more difficult. Hair, fur, thin wires, and motion blur can blend foreground and background elements together. The model has to measure how much of an edge pixel belongs to the subject and set its opacity accordingly. This is called image matting.","Generally, these skills are taught with different kinds of focused data. This creates a failure point. A poor training mix can produce unbalanced models where improvements in one skill come at the expense of the other. Outputs either miss parts of the subject or lack clean edges.","Real images are complex and require deftness in foreground recognition and boundary precision. This was our key insight when training FeyNoBg.","We chose BiRefNet as the foundation for FeyNoBg, as its architecture already matched our goal. BiRefNet gives two parts of the model complementary responsibilities. Its localization module finds the foreground, while its reconstruction module traces the subject’s boundary.","The model starts by passing the input image through a feature extractor that runs in four stages. Early stages capture local details. Later stages combine the gathered details into broader image representations called feature maps. The localization module uses these maps to find the subject, while the reconstruction module uses them to recover its boundary.","The third stage of the feature extractor has the hardest job. It sees enough of the image to reason about the whole subject while retaining the spatial detail needed to represent its shape. Both localization and boundary reconstruction depend heavily on the feature map produced here.","Given the rich image representation this stage holds, we expected that adding more depth here would help the model retain information better and thus improve performance. Accordingly, we expanded the third stage from 18 to 24 blocks. This grew the model modestly from 222M parameters to 263M.","We preserved every compatible pre-trained weight during the expansion. Only the six new blocks started untrained. This gave FeyNoBg additional capacity to learn new skills without forgetting what the base model already knows.","With room to learn more, it was time to teach an old model some new tricks.","Our first training run was with MaskFactory, a collection of synthetic image and mask pairs created for precise foreground segmentation. This was the only dataset used for the run.","If our intuition was correct, the resulting model would improve on some benchmarks while regressing on others. That is what happened in our controlled evaluation, where the model improved on the CAMO benchmark but regressed on DIS5K.","Next, we focused on building a more diverse dataset. We assembled 26.1K images from 10 datasets covering crowded scenes, camouflage, high resolution subjects, portraits, and anime, then trained for 7,000 steps. Our goal was to expose the model to as many scenarios as possible during training.","Synthetic scenes designed to help models generalize beyond familiar image collections.","We revised our original mix before the final run. We added 4,000 images from S3OD, reduced Anime to 500 images, and removed ThinObject-5K, HIM-2K, and COIFT. This left us with the sources that contributed the most useful, consistent training examples.","Combining the datasets introduced two problems. First, they varied greatly in size. Without limits, the largest sources would dominate training and cause the same specialization we saw before. Consequently, we capped each source to 4,000 images and then shuffled them together.","Second, the datasets used different annotations. Segmentation datasets provided foreground masks, while matting datasets provided alpha mattes. We converted both into binary foreground masks so every image shared the same training target.","The matting datasets therefore contributed precisely outlined subjects, not soft-opacity supervision.","This gave us one consistent training set with deliberately varied images. Our model could now learn to identify and outline foregrounds across a much wider range of scenarios.","To understand performance, we recorded S-measure. Scored from 0 to 1, S-measure compares the predicted foreground with the correct mask. It rewards both complete subjects and faithful shapes. A higher score is better.","We compared FeyNoBg’s S-Measure to the best published result on eight benchmarks covering camouflage, low contrast scenes, fine structures, high resolution images, and video. FeyNoBg led on four and was within 2% of the leader on the remaining four.","Notably, the broader training mix turned our DIS5K regression into a benchmark-leading result.","Image matting models are usually released as isolated repositories. Comparing models or fine-tuning one requires writing several adapters before any experiments can begin. This setup can quickly become too messy and frustrating to do good work in.","To solve this, we created NoBg. It is a Python library that provides a consistent interface to run and train background removal models. We used it ourselves to train FeyNoBg.","To encourage more development in this field, we are releasing NoBg in the open source. You can use it to run FeyNoBg, or train your own model.","A consistent interface should not cost performance. We compared NoBg’s BiRefNet with the original implementation at batch sizes 1, 2, and 4. NoBg delivered higher throughput, lower latency, and lower peak GPU memory at every size.","NoBg resizes and normalizes the image before inference. It then converts the model output into an alpha matte at the original size and saves the cutout as a transparent PNG.","NoBg provides the model, processor, and loss needed to train BiRefNet on your own image and mask pairs. It also works with the Hugging Face Trainer , which handles the training loop, checkpointing, and evaluation. Given a dataset with image and mask columns, training looks like this:","Download FeyNoBg from Hugging Face：https://huggingface.co/feyninc/FeyNobg and run the model with NoBg. Try it online in our Hugging Face Space：https://huggingface.co/spaces/feyninc/feynobg.","Install NoBg with pip install nobg . You can also find it on GitHub：https://github.com/feyninc/nobg.","FeyNoBg and NoBg are built by Feyn. Find us on X：x.com/feynai, GitHub：https://github.com/feyninc, or LinkedIn：https://www.linkedin.com/company/107081181/.","FeyNoBg builds on BiRefNet and the work of Peng Zheng, Dehong Gao, Deng-Ping Fan, Li Liu, Jorma Laaksonen, Wanli Ouyang, and Nicu Sebe. We also thank the teams that released the models, code, and datasets used here.","f(x) = argmin θ L(θ; your workflow) San Francisco, CA"],"articleImages":[{"sourceUrl":"https://usefeyn.com/blog/feynobg/feynobg-cover.png","alt":"FeyNoBg: A SOTA Model For Background Removal","afterParagraph":0,"url":"/media/articles/cms47q4jj00mxroep0x4at5sf/183bc571b8d28f89.png"},{"sourceUrl":"https://usefeyn.com/blog/feynobg/showcase/bikes-original.webp","alt":"Three rental bicycles parked on a busy city sidewalk","afterParagraph":0,"url":"/media/articles/cms47q4jj00mxroep0x4at5sf/613f59f61df67d43.webp"},{"sourceUrl":"https://usefeyn.com/blog/feynobg/showcase/bikes-cutout.webp","alt":"","afterParagraph":0,"url":"/media/articles/cms47q4jj00mxroep0x4at5sf/76df64af14de1f88.webp"},{"sourceUrl":"https://usefeyn.com/blog/feynobg/showcase/hair-original.webp","alt":"A person with long hair blowing across their face","afterParagraph":0,"url":"/media/articles/cms47q4jj00mxroep0x4at5sf/16aaab5304c4ed74.webp"},{"sourceUrl":"https://usefeyn.com/blog/feynobg/showcase/hair-cutout.webp","alt":"","afterParagraph":0,"url":"/media/articles/cms47q4jj00mxroep0x4at5sf/0def97284441c017.webp"},{"sourceUrl":"https://usefeyn.com/blog/feynobg/showcase/free-kick-original.webp","alt":"A football player taking a free kick against a defensive wall","afterParagraph":0,"url":"/media/articles/cms47q4jj00mxroep0x4at5sf/e36b62471d288d29.webp"}],"mediaStatus":"ok","articleBodyZh":["我们正在推出FeyNoBg，一种用于自动去除背景的最先进模型。在八个基准测试中，它在四个基准上获得了最佳已发表的S-measure，其余基准也只比领先者低不到2%。","在超高分辨率图像中测试显著物体的掩码，包括4K和8K场景。","我们还发布了NoBg，这是我们用来训练模型的库。使用NoBg可以运行FeyNoBg或训练您自己的背景去除模型。","两者都是开源的。在Hugging Face下载FeyNoBg：https://huggingface.co/feyninc/FeyNobg，或在GitHub上用NoBg构建：https://github.com/feyninc/nobg。","在您的计算机中，图像存储为像素网格。背景去除算法的目标是为每个像素预测一个不透明度值，使背景像素变为透明，前景像素保持不透明，边界像素变为半透明。","生成这种不透明度图需要两项技能。首先，模型必须将前景与背景分开。当主体站在简单的背景前时，可以通过比较颜色轻松完成。然而，在低对比度、拥挤或伪装的图像中，模型必须利用形状、纹理和上下文来识别哪些像素构成主体。","其次，模型必须描绘前景的边界。在简单图像中，颜色或纹理的明显变化提供了强边缘信号。但真实边界更难处理。头发、毛发、细线和运动模糊可以将前景和背景元素混合在一起。模型必须测量边缘像素属于主体的程度，并相应地设置其不透明度。这被称为图像抠图。","通常，这些技能通过不同类型的专注数据来教授。这会造成一个失效点。训练数据混合不当可能产生不平衡模型，使某一技能的提升以牺牲另一技能为代价。输出要么缺失主体部分，要么边缘不干净。","真实图像复杂，需要在前景识别和边界精度上都很熟练。这是我们在训练FeyNoBg时的关键洞察。","我们选择 BiRefNet 作为 FeyNoBg 的基础，因为其架构已经符合我们的目标。BiRefNet 为模型的两个部分赋予了互补的职责。其定位模块用于寻找前景，而重建模块用于追踪主体的边界。","模型首先将输入图像通过四个阶段运行的特征提取器。早期阶段捕捉局部细节。后期阶段将收集到的细节合并成更广泛的图像表示，称为特征图。定位模块使用这些特征图来寻找主体，而重建模块使用这些特征图来恢复其边界。","特征提取器的第三阶段承担着最艰巨的任务。它看到图像的足够部分以推断整个主体，同时保留表示其形状所需的空间细节。定位和边界重建都严重依赖此处产生的特征图。","鉴于该阶段持有的丰富图像表示，我们预计在此阶段增加更多深度将有助于模型更好地保留信息，从而提高性能。因此，我们将第三阶段从 18 个块扩展到 24 个块。模型由此从 2.22 亿参数适度增长到 2.63 亿。","在扩展过程中，我们保留了所有兼容的预训练权重。只有六个新区块是未训练的。这为 FeyNoBg 提供了额外的能力，以学习新技能而不遗忘基础模型已有的知识。","有了更多的学习空间，是时候教旧模型一些新技巧了。","我们的第一次训练运行使用了 MaskFactory，这是一个为精确前景分割创建的合成图像和掩码对集合。这是该次训练中使用的唯一数据集。","如果我们的直觉是正确的，生成的模型将在某些基准上提升，而在其他基准上退步。在我们的受控评估中确实如此，模型在 CAMO 基准上有所提升，但在 DIS5K 上有所退步。","接下来，我们专注于构建更加多样化的数据集。我们从 10 个数据集中组装了 26.1K 张图像，涵盖拥挤场景、伪装、高分辨率主体、人像和动漫，然后训练了 7,000 步。我们的目标是在训练期间尽可能多地让模型接触各种场景。","合成场景旨在帮助模型在熟悉的图像集合之外进行泛化。","在最终运行前，我们修改了最初的混合。我们添加了来自 S3OD 的 4,000 张图像，将 Anime 减少到 500 张，并移除了 ThinObject-5K、HIM-2K 和 COIFT。这样我们只保留了贡献最有用、最一致的训练样本的来源。","合并数据集引入了两个问题。首先，它们的规模差异很大。如果不加限制，最大的来源会主导训练，并导致我们之前看到的相同专化问题。因此，我们将每个来源限制为 4,000 张图像，然后将它们混合打乱。","其次，这些数据集使用了不同的标注。分割数据集提供前景掩码，而抠图数据集提供 alpha 通道。我们将两者都转换为二值前景掩码，以便每张图像都有相同的训练目标。","因此，抠图数据集贡献的是精确勾勒的主体，而不是软透明度的监督。","这为我们提供了一个一致的训练集，并故意包含各种不同的图像。我们的模型现在可以学习在更广泛的场景中识别和勾勒前景。","为了评估性能，我们记录了 S-measure。S-measure 从 0 到 1 评分，比较预测前景与正确掩码的匹配程度。它奖励完整的主体和准确的形状。分数越高越好。","我们将 FeyNoBg 的 S-measure 与八个基准上已发表的最佳结果进行了比较，这些基准包括伪装、低对比度场景、细结构、高分辨率图像和视频。FeyNoBg 在四个基准上排名第一，在其余四个基准上距离第一名不超过 2%。","值得注意的是，更广泛的训练混合把我们的 DIS5K 回归结果变成基准领先的成果。","图像抠图模型通常以独立仓库的形式发布。比较模型或对一个模型进行微调，需要在进行实验前编写多个适配器。这种设置很快就会变得混乱和令人沮丧，难以进行高质量工作。","为了解决这个问题，我们创建了 NoBg。它是一个 Python 库，提供一致的接口来运行和训练去背景模型。我们自己也使用它训练了 FeyNoBg。","为了鼓励这一领域的更多发展，我们将 NoBg 以开源形式发布。你可以使用它来运行 FeyNoBg，或训练你自己的模型。","一致的接口不应以性能为代价。我们在批量大小为 1、2 和 4 的情况下，比较了 NoBg 的 BiRefNet 与原始实现。NoBg 在每种批量大小下都提供了更高的吞吐量、更低的延迟和更低的峰值 GPU 内存使用量。","NoBg 在推理前会调整图像大小并进行归一化。然后，它将模型输出转换为原始尺寸的 alpha 通道，并将抠图结果保存为透明 PNG。","NoBg 提供了训练 BiRefNet 所需的模型、处理器和损失函数，可用于你自己的图像与掩码对。它也可以与 Hugging Face Trainer 一起使用，该工具可处理训练循环、检查点保存和评估。给定一个包含图像和掩码列的数据集，训练过程如下：","从 Hugging Face 下载 FeyNoBg：https://huggingface.co/feyninc/FeyNobg 并用 NoBg 运行模型。也可以在我们的 Hugging Face Space 在线尝试：https://huggingface.co/spaces/feyninc/feynobg。","使用 pip 安装 NoBg：pip install nobg。你也可以在 GitHub 上找到它：https://github.com/feyninc/nobg。","FeyNoBg 和 NoBg 由 Feyn 构建。找到我们：X 上 x.com/feynai，GitHub 上 https://github.com/feyninc，或 LinkedIn 上 https://www.linkedin.com/company/107081181/。","FeyNoBg 基于 BiRefNet 以及 Peng Zheng、Dehong Gao、Deng-Ping Fan、Li Liu、Jorma Laaksonen、Wanli Ouyang 和 Nicu Sebe 的工作构建。我们也感谢发布此处使用的模型、代码和数据集的团队。","f(x) = argmin θ L(θ; your workflow) 美国，加利福尼亚州，旧金山"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Feyn Labs 发布开源背景去除模型 FeyNoBg，并同步开源训练库 NoBg。材料称，该模型在八项基准中的四项取得最佳 S-measure，其余四项与领先者差距不超过 2%。","background":"FeyNoBg 基于 BiRefNet 架构，参数量由 222M 扩展至 263M，面向包括 4K、8K 场景在内的超高分辨率显著性目标抠图。背景去除同时涉及前景分离与边界透明度预测。","viewpoint":"Aioga 判断，FeyNoBg 的看点不只在基准成绩，也在于模型与训练库同时开源，降低了复现和训练同类模型的门槛。不过，现有材料未说明测试数据、硬件成本及实际部署表现，S-measure 不能单独代表全部使用体验。","implications":"如果相关基准结果能够被独立复核，FeyNoBg 可能为高分辨率图像抠图提供新的开源选项。其对毛发、柔边、低对比度和遮挡场景的实际处理质量，仍值得结合具体样例与更多指标观察。","nextStep":"建议进一步核对八项基准的名称、测试设置、对比模型和完整分数，并在 4K、8K、复杂背景及细小边界样本上进行复测。同时评估 NoBg 的训练流程、推理资源需求与许可证条件，再判断其是否适合生产使用。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-07-28T06:29:41.423Z","sourceHash":"6245a915bb98dbdc","review":{"approved":true,"groundedness":94,"clarity":91,"duplicationRisk":12,"blockingIssues":[],"notes":["“降低了复现和训练同类模型的门槛”属于合理推断，且已明确标注为“Aioga 判断”，没有冒充来源事实。","可考虑将“超高分辨率显著性目标抠图”改为“面向包括 4K、8K 场景在内的超高分辨率图像显著目标掩码预测与背景去除”，以更准确地区分显著目标检测、背景去除和图像抠图概念。","“遮挡场景”并未在来源材料中明确出现，但这里只是作为后续测试建议，不构成事实性断言。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","low-source-overlap","no-html","independent-ai-review"]}},"tags":["模型更新","Hacker News 热门（buzzing.cc 中文翻译）"],"translations":{"zh-CN":{"title":"FeyNoBg 发布：开源自动背景去除模型，在四项基准上达到 SOTA","summary":"Feyn Labs 推出 FeyNoBg，一个用于自动背景去除的 SOTA 模型。它在八个基准测试中的四项上取得最佳 S-measure 分数，其余四项与领先者差距在 2% 以内。该模型基于 BiRefNet 架构，参数量从 222M 扩展至 263M，同时开源了训练库 NoBg，模型和代码分别可在 Hugging Face 和 GitHub 获取。","category":"模型更新","source":"usefeyn.com","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg 发布：开源自动背景去除模型，在四项基准上达到 SOTA - Aioga AI资讯","description":"Feyn Labs 推出 FeyNoBg，一个用于自动背景去除的 SOTA 模型。它在八个基准测试中的四项上取得最佳 S-measure 分数，其余四项与领先者差距在 2% 以内。该模型基于 BiRefNet 架构，参数量从 222M 扩展至 263M，同时开源了训练库 NoBg，模型和代码分别可在 Hugging Face 和 GitHub 获取。","url":"https://www.aioga.com/news/cms47q4jj00mxroep0x4at5sf/"},"en":{"title":"FeyNoBg Release: Open-source automatic background removal model achieves SOTA on four benchmarks","summary":"Feyn Labs has launched FeyNoBg, a SOTA model for automatic background removal. It achieved the best S-measure scores on four out of eight benchmarks, with the remaining four within a 2% gap of the leaders. The model is based on the BiRefNet architecture, with parameters expanded from 222M to 263M. At the same time, the training library NoBg has been open-sourced, and the model and code can be accessed on Hugging Face and GitHub, respectively.","category":"Models","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg Release: Open-source automatic background removal model achieves SOTA on four benchmarks - Aioga AI News","description":"Feyn Labs has launched FeyNoBg, a SOTA model for automatic background removal. It achieved the best S-measure scores on four out of eight benchmarks, with the remaining four within...","url":"https://www.aioga.com/en/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:42:58.229Z"},"ja":{"title":"FeyNoBg 公開：オープンソースの自動背景除去モデルで、4つのベンチマークでSOTAを達成","summary":"Feyn Labs は FeyNoBg を発表しました。これは自動背景除去のための最先端（SOTA）モデルです。8つのベンチマークテストのうち4つで最高の S-measure スコアを獲得し、残りの4つでもトップとの差は 2% 以内です。このモデルは BiRefNet アーキテクチャに基づいており、パラメータ数は 222M から 263M に拡張されました。同時に、トレーニングライブラリ NoBg もオープンソース化されており、モデルとコードはそれぞれ Hugging Face と GitHub で入手可能です。","category":"モデル更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg 公開：オープンソースの自動背景除去モデルで、4つのベンチマークでSOTAを達成 - Aioga AIニュース","description":"Feyn Labs は FeyNoBg を発表しました。これは自動背景除去のための最先端（SOTA）モデルです。8つのベンチマークテストのうち4つで最高の S-measure スコアを獲得し、残りの4つでもトップとの差は 2% 以内です。このモデルは BiRefNet アーキテクチャに基づいており、パラメータ数は 222M から 263M に拡張されました。...","url":"https://www.aioga.com/ja/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:43:04.906Z"},"ko":{"title":"FeyNoBg 발표: 오픈소스 자동 배경 제거 모델, 네 가지 벤치마크에서 SOTA 달성","summary":"Feyn Labs는 자동 배경 제거용 SOTA 모델인 FeyNoBg를 출시했습니다. 이 모델은 8개의 벤치마크 중 4개에서 최고의 S-measure 점수를 기록했으며, 나머지 4개에서는 선도 모델과의 차이가 2% 이내였습니다. 이 모델은 BiRefNet 아키텍처를 기반으로 하며, 파라미터 수는 222M에서 263M로 확장되었고, 학습 라이브러리 NoBg도 오픈소스로 공개되었습니다. 모델과 코드는 각각 Hugging Face와 GitHub에서 확인할 수 있습니다.","category":"모델 업데이트","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg 발표: 오픈소스 자동 배경 제거 모델, 네 가지 벤치마크에서 SOTA 달성 - Aioga AI 뉴스","description":"Feyn Labs는 자동 배경 제거용 SOTA 모델인 FeyNoBg를 출시했습니다. 이 모델은 8개의 벤치마크 중 4개에서 최고의 S-measure 점수를 기록했으며, 나머지 4개에서는 선도 모델과의 차이가 2% 이내였습니다. 이 모델은 BiRefNet 아키텍처를 기반으로 하며, 파라미터 수는 222M에서 263M로 확장...","url":"https://www.aioga.com/ko/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:43:49.974Z"},"es":{"title":"FeyNoBg publicado: modelo de eliminación automática de fondo de código abierto, alcanzando SOTA en cuatro benchmarks","summary":"Feyn Labs ha lanzado FeyNoBg, un modelo SOTA para la eliminación automática de fondos. Obtuvo la mejor puntuación S-measure en cuatro de ocho puntos de referencia, y la diferencia con los líderes en los otros cuatro es inferior al 2%. El modelo se basa en la arquitectura BiRefNet, con parámetros que van de 222M a 263M, y al mismo tiempo se ha abierto la biblioteca de entrenamiento NoBg; el modelo y el código están disponibles en Hugging Face y GitHub, respectivamente.","category":"Modelos","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg publicado: modelo de eliminación automática de fondo de código abierto, alcanzando SOTA en cuatro benchmarks - Aioga Noticias de IA","description":"Feyn Labs ha lanzado FeyNoBg, un modelo SOTA para la eliminación automática de fondos. Obtuvo la mejor puntuación S-measure en cuatro de ocho puntos de referencia, y la diferencia...","url":"https://www.aioga.com/es/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:43:41.591Z"},"fr":{"title":"FeyNoBg publié : modèle open source de suppression automatique de l'arrière-plan, atteignant le SOTA sur quatre benchmarks","summary":"Feyn Labs a lancé FeyNoBg, un modèle SOTA pour la suppression automatique de l'arrière-plan. Il a obtenu le meilleur score S-measure sur quatre des huit benchmarks, tandis que pour les quatre autres, l'écart avec les leaders reste inférieur à 2 %. Le modèle est basé sur l'architecture BiRefNet, avec un nombre de paramètres passant de 222M à 263M, et la bibliothèque d'entraînement NoBg a également été open source. Le modèle et le code sont disponibles respectivement sur Hugging Face et GitHub.","category":"Modèles","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg publié : modèle open source de suppression automatique de l'arrière-plan, atteignant le SOTA sur quatre benchmarks - Aioga Actualités IA","description":"Feyn Labs a lancé FeyNoBg, un modèle SOTA pour la suppression automatique de l'arrière-plan. Il a obtenu le meilleur score S-measure sur quatre des huit benchmarks, tandis que pour...","url":"https://www.aioga.com/fr/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:44:24.052Z"},"de":{"title":"FeyNoBg veröffentlicht: Open-Source-Modell zur automatischen Hintergrundentfernung, erreicht SOTA bei vier Benchmark-Aufgaben","summary":"Feyn Labs hat FeyNoBg vorgestellt, ein SOTA-Modell zur automatischen Hintergrundentfernung. Es erzielte in vier von acht Benchmark-Tests die besten S-Measure-Werte, in den verbleibenden vier lag der Abstand zu den führenden Modellen bei weniger als 2%. Das Modell basiert auf der BiRefNet-Architektur, die Anzahl der Parameter wurde von 222M auf 263M erweitert, und gleichzeitig wurde die Trainingsbibliothek NoBg Open Source gestellt. Das Modell und der Code sind jeweils auf Hugging Face und GitHub verfügbar.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg veröffentlicht: Open-Source-Modell zur automatischen Hintergrundentfernung, erreicht SOTA bei vier Benchmark-Aufgaben - Aioga KI-News","description":"Feyn Labs hat FeyNoBg vorgestellt, ein SOTA-Modell zur automatischen Hintergrundentfernung. Es erzielte in vier von acht Benchmark-Tests die besten S-Measure-Werte, in den verbleib...","url":"https://www.aioga.com/de/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:44:26.557Z"},"pt-BR":{"title":"FeyNoBg publicado: modelo de remoção automática de fundo de código aberto, atingindo SOTA em quatro benchmarks","summary":"Feyn Labs lançou o FeyNoBg, um modelo SOTA para remoção automática de fundo. Ele alcançou a melhor pontuação S-measure em quatro dos oito benchmarks, com a diferença em relação ao líder nas quatro demais sendo inferior a 2%. O modelo é baseado na arquitetura BiRefNet, com o número de parâmetros variando de 222M a 263M, e também disponibilizou a biblioteca de treinamento NoBg de código aberto. O modelo e o código podem ser encontrados no Hugging Face e no GitHub, respectivamente.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg publicado: modelo de remoção automática de fundo de código aberto, atingindo SOTA em quatro benchmarks - Aioga Notícias de IA","description":"Feyn Labs lançou o FeyNoBg, um modelo SOTA para remoção automática de fundo. Ele alcançou a melhor pontuação S-measure em quatro dos oito benchmarks, com a diferença em relação ao...","url":"https://www.aioga.com/pt-BR/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:45:05.603Z"},"ru":{"title":"FeyNoBg выпущен: открытая модель для автоматического удаления фона, достигает SOTA по четырем эталонам","summary":"Feyn Labs выпустила FeyNoBg, SOTA-модель для автоматического удаления фона. Она достигла наилучшего показателя S-measure по четырем из восьми бенчмарков, а по оставшимся четырем отставание от лидеров составляет менее 2%. Модель основана на архитектуре BiRefNet, число параметров увеличилось с 222M до 263M, одновременно была открыта библиотека для обучения NoBg. Модель и код доступны на Hugging Face и GitHub соответственно.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg выпущен: открытая модель для автоматического удаления фона, достигает SOTA по четырем эталонам - Aioga Новости ИИ","description":"Feyn Labs выпустила FeyNoBg, SOTA-модель для автоматического удаления фона. Она достигла наилучшего показателя S-measure по четырем из восьми бенчмарков, а по оставшимся четырем от...","url":"https://www.aioga.com/ru/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:45:06.818Z"},"ar":{"title":"FeyNoBg نشرت: نموذج مفتوح المصدر لإزالة الخلفية تلقائيًا، وصل إلى أفضل أداء على أربع مجموعات معيارية","summary":"أطلقت Feyn Labs نموذج FeyNoBg، وهو نموذج حديث لإزالة الخلفية تلقائيًا. قد حقق أفضل درجات S-measure في أربعة من ثمانية اختبارات معيارية، والفجوة مع المتصدرين في الأربعة الأخرى لا تتجاوز 2%. يعتمد النموذج على بنية BiRefNet، وتم توسيع عدد المعلمات من 222 مليون إلى 263 مليون، كما تم إطلاق مكتبة التدريب NoBg كمصدر مفتوح، ويمكن الحصول على النموذج والكود على Hugging Face وGitHub على التوالي.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg نشرت: نموذج مفتوح المصدر لإزالة الخلفية تلقائيًا، وصل إلى أفضل أداء على أربع مجموعات معيارية - Aioga أخبار الذكاء الاصطناعي","description":"أطلقت Feyn Labs نموذج FeyNoBg، وهو نموذج حديث لإزالة الخلفية تلقائيًا. قد حقق أفضل درجات S-measure في أربعة من ثمانية اختبارات معيارية، والفجوة مع المتصدرين في الأربعة الأخرى لا تت...","url":"https://www.aioga.com/ar/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:45:59.586Z"},"hi":{"title":"FeyNoBg प्रकाशित: ओपन-सोर्स स्वचालित पृष्ठभूमि हटाने वाला मॉडल, चार मानकों पर SOTA तक पहुँच गया","summary":"Feyn Labs ने FeyNoBg पेश किया, जो एक स्वचालित पृष्ठभूमि हटाने के लिए SOTA मॉडल है। इसने आठ बेंचमार्क टेस्ट में से चार में सर्वोत्तम S-माप अंक प्राप्त किए, जबकि बाकी चार में अग्रणी से अंतर 2% के भीतर रहा। यह मॉडल BiRefNet वास्तुकला पर आधारित है, जिसके पैरामीटर 222M से बढ़ाकर 263M कर दिए गए हैं, और साथ ही प्रशिक्षण पुस्तकालय NoBg को ओपन सोर्स किया गया है। मॉडल और कोड क्रमशः Hugging Face और GitHub पर उपलब्ध हैं।","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg प्रकाशित: ओपन-सोर्स स्वचालित पृष्ठभूमि हटाने वाला मॉडल, चार मानकों पर SOTA तक पहुँच गया - Aioga AI समाचार","description":"Feyn Labs ने FeyNoBg पेश किया, जो एक स्वचालित पृष्ठभूमि हटाने के लिए SOTA मॉडल है। इसने आठ बेंचमार्क टेस्ट में से चार में सर्वोत्तम S-माप अंक प्राप्त किए, जबकि बाकी चार में अग्रणी...","url":"https://www.aioga.com/hi/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:45:46.271Z"},"it":{"title":"FeyNoBg rilasciato: modello open source per la rimozione automatica dello sfondo, raggiunge SOTA su quattro benchmark","summary":"Feyn Labs ha lanciato FeyNoBg, un modello SOTA per la rimozione automatica dello sfondo. Ha ottenuto il punteggio S-measure migliore in quattro degli otto benchmark, mentre per gli altri quattro la differenza con i leader è entro il 2%. Il modello si basa sull'architettura BiRefNet, con un numero di parametri che va da 222M a 263M, ed è stata resa open source la libreria di addestramento NoBg; il modello e il codice sono disponibili rispettivamente su Hugging Face e GitHub.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg rilasciato: modello open source per la rimozione automatica dello sfondo, raggiunge SOTA su quattro benchmark - Aioga Notizie IA","description":"Feyn Labs ha lanciato FeyNoBg, un modello SOTA per la rimozione automatica dello sfondo. Ha ottenuto il punteggio S-measure migliore in quattro degli otto benchmark, mentre per gli...","url":"https://www.aioga.com/it/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:46:40.115Z"},"nl":{"title":"FeyNoBg uitgebracht: open-source model voor automatische achtergrondverwijdering, bereikt SOTA op vier benchmarks","summary":"Feyn Labs heeft FeyNoBg gelanceerd, een SOTA-model voor automatische achtergrondverwijdering. Het behaalde de hoogste S-measure-score op vier van de acht benchmarks, terwijl het verschil met de leiders bij de overige vier benchmarks minder dan 2% bedraagt. Het model is gebaseerd op de BiRefNet-architectuur, met parameters die zijn uitgebreid van 222M naar 263M, en tegelijkertijd is de trainingsbibliotheek NoBg open source gemaakt. Het model en de code zijn respectievelijk beschikbaar op Hugging Face en GitHub.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg uitgebracht: open-source model voor automatische achtergrondverwijdering, bereikt SOTA op vier benchmarks - Aioga AI-nieuws","description":"Feyn Labs heeft FeyNoBg gelanceerd, een SOTA-model voor automatische achtergrondverwijdering. Het behaalde de hoogste S-measure-score op vier van de acht benchmarks, terwijl het ve...","url":"https://www.aioga.com/nl/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:46:41.149Z"},"tr":{"title":"FeyNoBg Yayınladı: Açık kaynaklı otomatik arka plan kaldırma modeli, dört benchmarkta SOTA'ya ulaştı","summary":"Feyn Labs, otomatik arka plan kaldırma için SOTA bir model olan FeyNoBg'yi tanıttı. Sekiz kıyaslama testinden dördünde en iyi S-measure puanını elde etti, diğer dört testte ise liderle farkı %2'nin altında kaldı. Model, BiRefNet mimarisi üzerine kuruludur ve parametre sayısı 222M'den 263M'ye yükseltilmiştir; aynı zamanda eğitim kütüphanesi NoBg de açık kaynak olarak sunulmuştur. Model ve koda sırasıyla Hugging Face ve GitHub üzerinden erişilebilir.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg Yayınladı: Açık kaynaklı otomatik arka plan kaldırma modeli, dört benchmarkta SOTA'ya ulaştı - Aioga AI Haberleri","description":"Feyn Labs, otomatik arka plan kaldırma için SOTA bir model olan FeyNoBg'yi tanıttı. Sekiz kıyaslama testinden dördünde en iyi S-measure puanını elde etti, diğer dört testte ise lid...","url":"https://www.aioga.com/tr/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:47:21.203Z"},"vi":{"title":"FeyNoBg phát hành: Mô hình loại bỏ nền tự động mã nguồn mở, đạt SOTA trên bốn chuẩn đánh giá","summary":"Feyn Labs ra mắt FeyNoBg, một mô hình SOTA dùng để loại bỏ nền tự động. Nó đạt điểm S-measure cao nhất trong bốn trên tám bài kiểm tra chuẩn, bốn bài còn lại chỉ chênh lệch dưới 2% so với người dẫn đầu. Mô hình này dựa trên kiến trúc BiRefNet, số lượng tham số mở rộng từ 222M lên 263M, đồng thời phát hành công khai thư viện huấn luyện NoBg; mô hình và mã nguồn có thể được lấy trên Hugging Face và GitHub.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg phát hành: Mô hình loại bỏ nền tự động mã nguồn mở, đạt SOTA trên bốn chuẩn đánh giá - Tin tức AI Aioga","description":"Feyn Labs ra mắt FeyNoBg, một mô hình SOTA dùng để loại bỏ nền tự động. Nó đạt điểm S-measure cao nhất trong bốn trên tám bài kiểm tra chuẩn, bốn bài còn lại chỉ chênh lệch dưới 2%...","url":"https://www.aioga.com/vi/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:47:19.631Z"},"id":{"title":"FeyNoBg Rilis: Model penghapus latar belakang otomatis sumber terbuka, mencapai SOTA pada empat tolok ukur","summary":"Feyn Labs meluncurkan FeyNoBg, sebuah model SOTA untuk penghapusan latar belakang otomatis. Model ini memperoleh skor S-measure terbaik pada empat dari delapan tolok ukur, sedangkan empat lainnya hanya berbeda kurang dari 2% dari pemimpin. Model ini berbasis arsitektur BiRefNet, dengan jumlah parameter meningkat dari 222M menjadi 263M, dan juga merilis perpustakaan pelatihan NoBg secara terbuka. Model dan kodenya masing-masing dapat diakses di Hugging Face dan GitHub.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg Rilis: Model penghapus latar belakang otomatis sumber terbuka, mencapai SOTA pada empat tolok ukur - Berita AI Aioga","description":"Feyn Labs meluncurkan FeyNoBg, sebuah model SOTA untuk penghapusan latar belakang otomatis. Model ini memperoleh skor S-measure terbaik pada empat dari delapan tolok ukur, sedangka...","url":"https://www.aioga.com/id/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:48:01.538Z"},"th":{"title":"FeyNoBg เผยแพร่: โมเดลลบพื้นหลังอัตโนมัติแบบเปิดเผยซอร์ส ทำคะแนน SOTA ในมาตรฐานสี่รายการ","summary":"Feyn Labs เปิดตัว FeyNoBg ซึ่งเป็นโมเดล SOTA สำหรับการลบพื้นหลังอัตโนมัติ โมเดลนี้ทำคะแนน S-measure สูงสุดในสี่จากแปดเบนช์มาร์ก ที่เหลืออีกสี่คะแนนต่างจากผู้นำไม่เกิน 2% โมเดลนี้สร้างขึ้นบนสถาปัตยกรรม BiRefNet โดยพารามิเตอร์เพิ่มจาก 222M เป็น 263M พร้อมเปิดโค้ดห้องสมุดการฝึก NoBg แบบโอเพ่นซอร์ส โมเดลและโค้ดสามารถเข้าถึงได้ที่ Hugging Face และ GitHub ตามลำดับ","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg เผยแพร่: โมเดลลบพื้นหลังอัตโนมัติแบบเปิดเผยซอร์ส ทำคะแนน SOTA ในมาตรฐานสี่รายการ - ข่าว AI Aioga","description":"Feyn Labs เปิดตัว FeyNoBg ซึ่งเป็นโมเดล SOTA สำหรับการลบพื้นหลังอัตโนมัติ โมเดลนี้ทำคะแนน S-measure สูงสุดในสี่จากแปดเบนช์มาร์ก ที่เหลืออีกสี่คะแนนต่างจากผู้นำไม่เกิน 2% โมเดลนี้สร...","url":"https://www.aioga.com/th/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:48:04.800Z"},"pl":{"title":"FeyNoBg wydanie: Otwarty model automatycznego usuwania tła, osiągający SOTA w czterech benchmarkach","summary":"Feyn Labs wprowadza FeyNoBg, model SOTA do automatycznego usuwania tła. Osiągnął najwyższy wynik S-measure w czterech z ośmiu testów benchmarkowych, a w pozostałych czterech różnica w stosunku do liderów wynosi poniżej 2%. Model oparty jest na architekturze BiRefNet, a liczba parametrów rozszerzyła się z 222M do 263M. Jednocześnie udostępniono otwartoźródłową bibliotekę treningową NoBg, a model i kod są dostępne odpowiednio na Hugging Face i GitHub.","category":"模型更新","source":"Hacker News 热门（buzzing.cc 中文翻译）","aggregationSource":"Hacker News 热门（buzzing.cc 中文翻译）","pageTitle":"FeyNoBg wydanie: Otwarty model automatycznego usuwania tła, osiągający SOTA w czterech benchmarkach - Aioga Wiadomości AI","description":"Feyn Labs wprowadza FeyNoBg, model SOTA do automatycznego usuwania tła. Osiągnął najwyższy wynik S-measure w czterech z ośmiu testów benchmarkowych, a w pozostałych czterech różnic...","url":"https://www.aioga.com/pl/news/cms47q4jj00mxroep0x4at5sf/","contentTranslated":true,"sourceHash":"8e85524c43d6147c","translatedAt":"2026-07-28T05:48:50.114Z"}}}}