Research map / Learning theory and algorithms / Clustering
ML-LocNet: Improving Object Localization with Multi-view Learning Network
Xiaopeng Zhang, Yang Yang, Jiashi Feng
This paper tried a common change: Add dropout
add region dropout in phase one · PASCAL VOC 2007 trainval (design study) · CorLoc (%): 57.1 -> 57.9 Improved
What 795 accepted papers report for this change
- 549 improved
- 87 no clear effect
- 12 mixed
- 147 hurt
- Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference ICML 2026 · apply layer dropout to whole transformer blocks instead of attention and FFN sub-blocks separately · 20 TPP val loss, 271M · val loss: 2.377 -> 2.367Improved
- Quantifying the Uncertainty of Foundation Models with Singular Value Ensembles ICML 2026 · use MC Dropout · vision and SST-2 · accuracy: 86.3 -> 76.6 on Flowers102Hurt
- Motion Planning in Compressed Representation Spaces ICML 2026 · add nested dropout during training · WOMD guided generation · goal FDE: 4.145 -> 1.104Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add multi-view feature concatenation in phase one (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 52.5 -> 57.1 | Improved |
| add region dropout in phase one (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 57.1 -> 57.9 | Improved |
| add plain Fast-RCNN-style second phase without multi-view (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 57.9 -> 58.2 | No clear effect |
| add multi-view co-training in phase two (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 57.9 -> 61.6 | Improved |
| add random mini-batch sampling per view (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 61.6 -> 62.8 | Improved |
| add weighted loss on mined instances (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 62.8 -> 63.7 | Improved |
| use concatenation instead of multiple view losses in phase one (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 54.6 -> 57.9 | Improved |
| use diverse view parameters (compressed) instead of identical uncompressed views, phase one (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 56.2 -> 57.9 | Improved |
| use diverse view parameters (compressed) instead of identical uncompressed views, phase two (ablation) | ML-LocNet (model S) · PASCAL VOC 2007 trainval (design study) | CorLoc (%): 58.9 -> 63.7 | Improved |
| use ML-LocNet-S instead of previous best ODGA-S with same model | ML-LocNet · PASCAL VOC 2007 trainval | CorLoc mAP (%): 55.6 -> 63.7 | Improved |
| use ML-LocNet-L+ (mean of three models) instead of OICR-L+ | ML-LocNet · PASCAL VOC 2007 trainval | CorLoc mAP (%): 64.3 -> 68.6 | Improved |
| use ML-LocNet-L+ instead of OICR-L+ | ML-LocNet · PASCAL VOC 2007 test | detection mAP (%): 47.0 -> 49.7 | Improved |
Closest papers by meaning
- Learning to Localize Objects with Noisy Labeled Instances (AAAI 2019 · similarity 0.89)
- Evaluation for Weakly Supervised Object Localization: Protocol, Metrics, and Datasets (TPAMI 2022 · similarity 0.89)
- Evaluating Weakly Supervised Object Localization Methods Right (CVPR 2020 · similarity 0.89)
- Pairwise Similarity Knowledge Transfer for Weakly Supervised Object Localization (ECCV 2020 · similarity 0.88)
- Rethinking the Route Towards Weakly Supervised Object Localization (CVPR 2020 · similarity 0.88)
- ContextLocNet: Context-Aware Deep Network Models for Weakly Supervised Localization (ECCV 2016 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
