NearID：基於近身份干擾項的身份表徵學習

摘要

在評估以身份識別為核心的任務（如個性化生成與圖像編輯）時，現有視覺編碼器會將物件身份與背景上下文相互纏結，導致表徵與評估指標不可靠。我們提出首個基於理論的框架，透過「近身份干擾項」解決此缺陷：將語意相似但不同的實例置於與參考圖像完全相同的背景中，從而消除上下文捷徑，使身份特徵成為唯一區分信號。基於此原理，我們發布NearID數據集（包含1.9萬個身份標識與31.6萬組背景匹配的干擾項），並提出嚴格的邊距評估機制。在此設定下，預訓練編碼器表現不佳——其樣本成功率（SSR，一種嚴格的基於邊距的身份區分指標）低至30.7%，且常將干擾項排序置於真實跨視圖匹配結果之上。為解決此問題，我們在凍結骨幹網絡的基礎上，透過雙層對比學習目標來訓練身份感知表徵，強化「同一身份＞近身份干擾項＞隨機負樣本」的層級關係。該方法將SSR提升至99.2%，增強部件級區分能力達28.0%，並在人類對齊的個性化基準DreamBench++中與人類判斷達成更高一致性。項目頁面：https://gorluxor.github.io/NearID/

English

When evaluating identity-focused tasks such as personalized generation and image editing, existing vision encoders entangle object identity with background context, leading to unreliable representations and metrics. We introduce the first principled framework to address this vulnerability using Near-identity (NearID) distractors, where semantically similar but distinct instances are placed on the exact same background as a reference image, eliminating contextual shortcuts and isolating identity as the sole discriminative signal. Based on this principle, we present the NearID dataset (19K identities, 316K matched-context distractors) together with a strict margin-based evaluation protocol. Under this setting, pre-trained encoders perform poorly, achieving Sample Success Rates (SSR), a strict margin-based identity discrimination metric, as low as 30.7% and often ranking distractors above true cross-view matches. We address this by learning identity-aware representations on a frozen backbone using a two-tier contrastive objective enforcing the hierarchy: same identity > NearID distractor > random negative. This improves SSR to 99.2%, enhances part-level discrimination by 28.0%, and yields stronger alignment with human judgments on DreamBench++, a human-aligned benchmark for personalization. Project page: https://gorluxor.github.io/NearID/

NearID：基於近身份干擾項的身份表徵學習

NearID: Identity Representation Learning via Near-identity Distractors

摘要

Support