Multi Scale Table Detection With Resolution Adaptation

via “multi-scale table detection with resolution adaptation”

object-detection model by undefined. 33,94,499 downloads.

Unique: Implements scale-aware NMS that considers detection confidence and scale context when merging overlapping boxes, preventing duplicate detections while preserving small-table detections that might be suppressed by naive coordinate-based NMS. The resolution adaptation uses aspect-ratio-preserving padding rather than stretching, maintaining table proportions.

vs others: More effective than single-scale detection for documents with mixed table sizes because transformer attention can capture multi-scale context; outperforms image pyramid approaches (like FPN) because it processes each scale independently and merges results, reducing false positives from scale confusion.

via “multi-scale feature pyramid detection across image resolutions”

object-detection model by undefined. 2,23,706 downloads.

Unique: YOLOv10 uses an improved PAN (Path Aggregation Network) with bidirectional feature fusion, enabling better information flow between scales compared to YOLOv8's simpler FPN, resulting in ~2-3% mAP improvement on small objects.

vs others: More efficient than Faster R-CNN's region proposal approach for multi-scale detection; simpler than cascade detectors (which require multiple stages) while achieving comparable accuracy on small objects.

via “multi-scale inference with dynamic input resolution”

object-detection model by undefined. 86,897 downloads.

Unique: YOLO11 inference pipeline automatically handles aspect-ratio-preserving letterboxing and coordinate transformation without explicit user code. Supports inference at any resolution; internally optimizes tensor shapes for GPU memory efficiency. Provides built-in multi-scale inference mode (runs model at 0.5x, 1.0x, 1.5x scales and merges results) accessible via single parameter.

vs others: More flexible than fixed-resolution detectors (Faster R-CNN typically requires 800x600 or similar); automatic coordinate transformation more robust than manual scaling; built-in multi-scale mode simpler than implementing custom tiling logic.

via “multi-scale feature extraction with feature pyramid network”

object-detection model by undefined. 1,06,918 downloads.

Unique: Combines FPN with deformable attention, where deformable modules adaptively sample features across FPN levels based on object location and scale. This enables scale-aware attention that standard FPN + fixed attention cannot achieve, improving detection of objects at extreme scales.

vs others: More effective than single-scale detection (standard YOLO) for scale-diverse datasets because FPN explicitly processes multiple scales, while remaining more efficient than naive multi-resolution inference that runs the full model multiple times.

via “hierarchical-multi-scale-feature-extraction”

* ⭐ 01/2022: [Patches Are All You Need (ConvMixer)](https://arxiv.org/abs/2201.09792)

Unique: Achieves multi-scale feature extraction through pure convolutional downsampling stages inspired by ViT hierarchical design, avoiding transformer-specific mechanisms while maintaining the ability to produce feature pyramids competitive with Swin Transformer's shifted-window hierarchical attention

vs others: Produces multi-scale features with lower computational overhead than Swin Transformer's windowed attention while maintaining competitive detection/segmentation performance on COCO and ADE20K benchmarks