mlsanityData to 14 Oct 2026

FrameLite: Decision-Oriented Aesthetic Cropping with a Compact Vision-Language Model

arXiv 2610.17639 ↗First posted 13 Oct 2026cs.CVcited by 0

On the map: Multimodal LLMs / vision-language models → Image editing and quality assessment

Abstract (arXiv)

Aesthetic image cropping improves composition by balancing subjects, context, and image boundaries. Large vision-language models bring semantic knowledge to this task, but their scale limits lightweight photography assistance. We propose FrameLite, a compact cropping model built on Qwen3.5-0.8B. Our approach is motivated by the observation that a small model can generate better crops than its first answer suggests. We develop three-stage training to turn this potential into stronger direct predictions. Spatial and composition training establishes localization and framing preferences. Quality-Guided Online Supervision (QOS) then learns the highest-quality complete answers among the current policy's generated candidates. Greedy-Referenced Optimization (GRO) further refines the policy using positive and negative quality differences relative to its current greedy answer. Both online stages use annotation-derived feedback, while inference returns one coordinate answer without search or an external scorer. Experiments show that FrameLite outperforms the compared large-VLM croppers across public benchmarks with a substantially smaller backbone. Controlled studies confirm that selecting high-quality training answers improves the final model beyond online sampling alone. We will release our code and model weights at https://github.com/JianzeLi-114/FrameLite .

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Image editing and quality assessment, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 27 of its 42 references that are in the map. First posted on arXiv 13 Oct 2026.

Builds on

27 of 42 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Venus: Benchmarking and Empowering Multimodal Large Language Models for Aesthetic Guidance and Cropping27 Feb
  2. 2025The Photographer Eye: Teaching Multimodal Large Language Models to Understand Image Aesthetics like Photographers23 Sep
  3. 2024AesExpert: Towards Multi-modality Foundation Model for Image Aesthetics Perception15 Apr
  4. 2023VILA: Learning Image Aesthetics from User Comments with Vision-Language Pretraining24 Mar

Other references · 23

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  3. 2016Modeling Context in Referring Expressionscited by 1,063
  4. 2023RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignmentcited by 686
  5. 2016Self-critical Sequence Training for Image Captioningcited by 547
  6. 2019Reliable and Efficient Image Cropping: A Grid Anchor based Approachcited by 33
  7. 2017Quantitative Analysis of Automatic Image Cropping Algorithms: A Dataset and Comparative Studycited by 27
  8. 2019Grid Anchor based Image Cropping: A New Benchmark and An Efficient Modelcited by 26
  9. 2017Learning to Compose with Professional Photographs on the Webcited by 25
  10. 2019Image Cropping with Composition and Saliency Aware Aesthetic Score Mapcited by 21
  11. 2017Deep Cropping via Attention Box Prediction and Aesthetics Assessmentcited by 17
  12. 2024Spatial-Semantic Collaborative Cropping for User Generated Contentcited by 13
11 more
  1. 2023Learning Subject-Aware Cropping by Outpainting Professional Photoscited by 12
  2. 2022Human-centric Image Cropping with Partition-aware and Content-preserving Featurescited by 11
  3. 2024Cropper: Vision-Language Model for Image Cropping through In-Context Learningcited by 11
  4. 2023Beyond Image Borders: Learning Feature Extrapolation for Unbounded Image Compositioncited by 10
  5. 2025PhotoFramer: Multi-modal Image Composition Instructioncited by 9
  6. 2025ProCrop: Learning Aesthetic Image Cropping from Professional Compositionscited by 8
  7. 2022ClipCrop: Conditioned Cropping Driven by Vision-Language Modelcited by 7
  8. 2025Towards Smart Point-and-Shoot Photographycited by 6
  9. 2026CROP: Expert-Aligned Image Cropping via Compositional Reasoning and Optimizing Preferencecited by 1
  10. 2026ShutterMuse: Capture-Time Photography Guidance with MLLMscited by 1
  11. 2026Global Attention-Fused Image Cropping with Attention-Guided and Global-Aligned Crop Evaluatorcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Discrete Annotation, Continuous Preference: Rethinking Supervision for Accurate and Generalizable Aesthetic Image Cropping
  2. 2026COMEX: A Composition-Grounded Benchmark and Learning Framework for Explainable Aesthetic Image Cropping
  3. 2025AesCrop: Aesthetic-driven Cropping Guided by Composition
  4. 2026Look Where It Matters: High-Resolution Crops Retrieval for Efficient VLMs
  5. 2025Global Compression Commander: Plug-and-Play Inference Acceleration for High-Resolution Large Vision-Language Models
  6. 2026Learning to Look Again: Loss-Gap Supervision for Free-form Crop Routing in Vision-Language Models
  7. 2026Learning to Focus and Precise Cropping: A Reinforcement Learning Framework with Information Gaps and Grounding Loss for MLLMs
  8. 2026Aesthetic Camera Viewpoint Suggestion with 3D Aesthetic Field