Research map / Architectures and efficiency / RNNs and normalization
[Re] Learning to count everything
Maša Kljun, Matija Teršek, Domen Vreš
Read the paperCheck your idea against 137,217 papers
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 |
|---|---|---|---|
| Use 2 exemplars instead of 1 (ablation) | FamNet · FSC-147 Val Set | MAE: 26.55 -> 24.09 | Improved |
| Use 2 exemplars instead of 1 (ablation) | FamNet · FSC-147 Val Set | RMSE: 77.01 -> 72.37 | Improved |
| Use 3 exemplars instead of 2 (ablation) | FamNet · FSC-147 Val Set | MAE: 24.09 -> 23.75 | Improved |
| Use 3 exemplars instead of 2 (ablation) | FamNet · FSC-147 Val Set | RMSE: 72.37 -> 69.07 | Improved |
| Add multi-scale image feature (ablation) | FamNet · FSC-147 Val Set | MAE: 32.70 -> 27.80 | Improved |
| Add multi-scale image feature (ablation) | FamNet · FSC-147 Val Set | RMSE: 104.31 -> 93.53 | Improved |
| Add multi-scale exemplar feature (ablation) | FamNet · FSC-147 Val Set | MAE: 27.80 -> 24.32 | Improved |
| Add multi-scale exemplar feature (ablation) | FamNet · FSC-147 Val Set | RMSE: 93.53 -> 70.94 | Improved |
| Add test-time adaptation (ablation) | FamNet with multi-scale image and exemplar features · FSC-147 Val Set | MAE: 24.32 -> 23.75 | Improved |
| Add test-time adaptation (ablation) | FamNet with multi-scale image and exemplar features · FSC-147 Val Set | RMSE: 70.94 -> 69.07 | Improved |
| Train FamNet on CARPK data | FamNet-- on CARPK · CARPK | MAE: 28.84 -> 18.19 | Improved |
| Train FamNet on CARPK data | FamNet-- on CARPK · CARPK | RMSE: 44.47 -> 33.66 | Improved |
Closest papers by meaning
- A Fixed-Point Approach to Unified Prompt-Based Counting (AAAI 2024 · similarity 0.76)
- Language Models Need Inductive Biases to Count Inductively (ICLR 2025 · similarity 0.75)
- Learning to Count without Annotations (CVPR 2024 · similarity 0.75)
- Counting in Small Transformers: The Delicate Interplay between Attention and Feed-Forward Layers (ICML 2025 · similarity 0.75)
- Light-IF: Endowing LLMs with Generalizable Reasoning via Preview and Self-Checking for Complex Instruction Following (AAAI 2026 · similarity 0.74)
- On a Combinatorial Problem Arising in Machine Teaching (ICML 2024 · similarity 0.74)
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.
