MAPS provides 2618 validated 3D meshes and a controllable rendering pipeline to attribute vision model recognition failures to specific scene parameters, finding camera distance and elevation as the dominant failure factors across 20 tested models.
arXiv preprint arXiv:2105.07197 , year=
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
verdicts
UNVERDICTED 7representative citing papers
A human-centered OOD spectrum based on perceptual difficulty shows vision-language models align best with human errors across regimes, with CNNs stronger on near-OOD and ViTs on far-OOD.
SPANetV2 is a vision backbone built around a new spectral-adaptive modulation mixer that outperforms prior models on ImageNet-1K classification, COCO detection, and ADE20K segmentation.
A Shapley-value method with interaction terms that explains object detector decisions by capturing collective pixel contributions for localization and classification.
LSTM-decoder VLMs reach 80-87% of human noise ceiling in fixation alignment versus 40-59% for Transformer decoders, though LSTM maps are diffuse and less task-specific while CNN-Transformer maps better predict synthetic neural responses.
TEXTER generates zero-shot textual explanations for image classifiers by isolating decision-critical features from contributing neurons, mapping them into CLIP space, and using sparse autoencoders for improved interpretability in Transformers.
Feedback Former improves cell image segmentation accuracy by feeding detailed feature maps back from near the output to lower transformer layers, outperforming non-feedback baselines with lower computational cost on three datasets.
citing papers explorer
-
MAPS: A Synthetic Dataset for Probing Vision Models in a Controlled 3D Scene Space
MAPS provides 2618 validated 3D meshes and a controllable rendering pipeline to attribute vision model recognition failures to specific scene parameters, finding camera distance and elevation as the dominant failure factors across 20 tested models.
-
Do Machines Fail Like Humans? A Human-Centred Out-of-Distribution Spectrum for Mapping Error Alignment
A human-centered OOD spectrum based on perceptual difficulty shows vision-language models align best with human errors across regimes, with CNNs stronger on near-OOD and ViTs on far-OOD.
-
Spectral-Adaptive Modulation Networks for Visual Perception
SPANetV2 is a vision backbone built around a new spectral-adaptive modulation mixer that outperforms prior models on ImageNet-1K classification, COCO detection, and ADE20K segmentation.
-
Explaining Object Detectors via Collective Contribution of Pixels
A Shapley-value method with interaction terms that explains object detector decisions by capturing collective pixel contributions for localization and classification.
-
Attention Alignment Between Humans and Vision-Language Models
LSTM-decoder VLMs reach 80-87% of human noise ceiling in fixation alignment versus 40-59% for Transformer decoders, though LSTM maps are diffuse and less task-specific while CNN-Transformer maps better predict synthetic neural responses.
-
Zero-Shot Textual Explanations via Translating Decision-Critical Features
TEXTER generates zero-shot textual explanations for image classifiers by isolating decision-critical features from contributing neurons, mapping them into CLIP space, and using sparse autoencoders for improved interpretability in Transformers.
-
Accuracy Improvement of Cell Image Segmentation Using Feedback Former
Feedback Former improves cell image segmentation accuracy by feeding detailed feature maps back from near the output to lower transformer layers, outperforming non-feedback baselines with lower computational cost on three datasets.