# Semantically Compatible Knowledge Distillation for Cross-Domain Object Detection with Vision Foundation Models

arXiv: https://arxiv.org/abs/2608.20916 · HTML: https://mlsanity.com/paper/2608.20916/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Detection and dense prediction distillation, one of 12 directions in Knowledge distillation.
- Builds on 26 of its 35 references that are in the map.
- First posted on arXiv 21 Aug 2026.

## Placed in

- Knowledge distillation / [Detection and dense prediction distillation](https://mlsanity.com/distill/detection-and-dense-prediction-distillation/)

## Builds on (most cited first, 26 of 26)

- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [Domain Adaptive Faster R-CNN for Object Detection in the Wild](https://mlsanity.com/paper/1803.03243/) (2018)
- [MIC: Masked Image Consistency for Context-Enhanced Domain Adaptation](https://mlsanity.com/paper/2212.01322/) (2022)
- [Channel-wise Knowledge Distillation for Dense Prediction](https://mlsanity.com/paper/2011.13256/) (2020)
- [Focal and Global Knowledge Distillation for Detectors](https://mlsanity.com/paper/2111.11837/) (2021)
- [Cross-Domain Adaptive Teacher for Object Detection](https://mlsanity.com/paper/2111.13216/) (2021)
- [CLIP the Gap: A Single Domain Generalization Approach for Object Detection](https://mlsanity.com/paper/2301.05499/) (2023)
- [One-for-All: Bridge the Gap Between Heterogeneous Architectures in Knowledge Distillation](https://mlsanity.com/paper/2310.19444/) (2023)
- [SIGMA: Semantic-complete Graph Matching for Domain Adaptive Object Detection](https://mlsanity.com/paper/2203.06398/) (2022)
- [PODA: Prompt-driven Zero-shot Domain Adaptation](https://mlsanity.com/paper/2212.03241/) (2022)
- [Learning Domain Adaptive Object Detection with Probabilistic Teacher](https://mlsanity.com/paper/2206.06293/) (2022)
- [Contrastive Mean Teacher for Domain Adaptive Object Detectors](https://mlsanity.com/paper/2305.03034/) (2023)
- [Cross Domain Object Detection by Target-Perceived Dual Branch Distillation](https://mlsanity.com/paper/2205.01291/) (2022)
- [Unified Language-driven Zero-shot Domain Adaptation](https://mlsanity.com/paper/2404.07155/) (2024)
- [Multi-Granularity Alignment Domain Adaptation for Object Detection](https://mlsanity.com/paper/2203.16897/) (2022)
- [Towards Robust Adaptive Object Detection under Noisy Annotations](https://mlsanity.com/paper/2204.02620/) (2022)
- [CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection](https://mlsanity.com/paper/2403.19278/) (2024)
- [DATR: Unsupervised Domain Adaptive Detection Transformer with Dataset-Level Adaptation and Prototypical Alignment](https://mlsanity.com/paper/2405.11765/) (2024)
- [Unsupervised Domain Adaptive Detection with Network Stability Analysis](https://mlsanity.com/paper/2308.08182/) (2023)
- [Large Self-Supervised Models Bridge the Gap in Domain Adaptive Object Detection](https://mlsanity.com/paper/2503.23220/) (2025)
- [Generalized Diffusion Detector: Mining Robust Features from Diffusion Models for Domain-Generalized Detection](https://mlsanity.com/paper/2503.02101/) (2025)
- [Differential Alignment for Domain Adaptive Object Detection](https://mlsanity.com/paper/2412.12830/) (2024)
- [Diffusion Domain Teacher: Diffusion Guided Domain Adaptive Object Detector](https://mlsanity.com/paper/2506.04211/) (2025)
- [UPRE: Zero-Shot Domain Adaptation for Object Detection via Unified Prompt and Representation Enhancement](https://mlsanity.com/paper/2507.00721/) (2025)
- [Boosting Domain Generalized and Adaptive Detection with Diffusion Models: Fitness, Generalization, and Transferability](https://mlsanity.com/paper/2506.21042/) (2025)

## Abstract (arXiv)

Vision foundation models (VFMs) offer strong generalization capabilities for domain-adaptive object detection (DAOD). However, existing VFM-based methods overlook the spatial-scale discrepancy between teacher and student feature maps, resulting in semantic incompatibility that weakens both feature alignment and pseudo-label learning. Moreover, domain shift can cause source-trained VFM teachers to miss target-domain objects, limiting the quality of their pseudo-labels. To address these issues, we propose the Semantic Localization-Enhanced Teacher (SLE-T), a semantically compatible knowledge-distillation framework built around a lightweight SLE Adapter for DINOv2. SLE Adapter injects pretrained local-texture priors into DINOv2 to improve cross-domain recognition and reformulates its features into dense representations that are spatially and semantically compatible with the student detector. SLE-T transfers the resulting teacher knowledge through either pseudo-label learning or feature alignment. We instantiate SLE-T with DINOv2-B and DINOv2-L (the ViT-B and ViT-L variants) and compare them with the larger DINOv2-G teacher. Extensive experiments on three DAOD benchmarks demonstrate that our method achieves state-of-the-art performance, and ablation studies confirm the importance of teacher-student semantic compatibility. Notably, SLE-T with DINOv2-B produces competitive or superior pseudo-labels using approximately one-quarter of the training time of DINOv2-G and substantially less GPU memory, demonstrating efficient VFM knowledge transfer under limited computational resources.