Object Detection In Aerial Images On Dota 1
评估指标
mAP
评测结果
各个模型在此基准测试上的表现结果
比较表格
| 模型名称 | mAP |
|---|---|
| eautodet-efficient-architecture-search-for | 77.05% |
| an-empirical-study-of-remote-sensing | 77.72% |
| mtp-advancing-remote-sensing-foundation-model | 80.77% |
| advancing-plain-vision-transformer-towards | 81.01% |
| advancing-plain-vision-transformer-towards | 81.24% |
| lsknet-a-foundation-lightweight-backbone-for | 81.37% |
| adaptive-rotated-convolution-for-rotated | 81.77% |
| cfc-net-a-critical-feature-capturing-network | 73.50% |
| decouplenet-a-lightweight-backbone-network | 78.04% |
| oriented-reppoints-for-aerial-object | 77.63% |
| redet-a-rotation-equivariant-detector-for | 80.10% |
| dynamic-anchor-learning-for-arbitrary | 76.95% |
| dynamic-refinement-network-for-oriented-and | 73.23% |
| oriented-r-cnn-for-object-detection | 80.87% |
| r2cnn-multi-dimensional-attention-based | 72.61% |
| adaptive-period-embedding-for-representing | 75.75 |
| lsknet-a-foundation-lightweight-backbone-for | 81.85% |
| learning-roi-transformer-for-detecting | 69.56% |
| beyond-bounding-box-convex-hull-feature | 76.67% |
| rethinking-rotated-object-detection-with | 80.23% |
| oriented-objects-as-pairs-of-middle-lines | 72.8% |
| mtp-advancing-remote-sensing-foundation-model | 80.67% |
| spatial-transform-decoupling-for-oriented | 82.24% |
| an-empirical-study-of-remote-sensing | 76.12% |
| pp-yoloe-r-an-efficient-anchor-free-rotated | 79.71% |
| strip-r-cnn-large-strip-convolution-for | 82.75% |
| a-general-gaussian-heatmap-labeling-for | 76.95% |
| align-deep-features-for-oriented-object | 79.42% |
| pp-yoloe-r-an-efficient-anchor-free-rotated | 80.02% |
| mocae-mixture-of-calibrated-experts | 82.62% |
| optimization-for-oriented-object-detection | 77.62% |
| gliding-vertex-on-the-horizontal-bounding-box | 75.02% |
| pp-yoloe-r-an-efficient-anchor-free-rotated | 80.73% |
| piou-loss-towards-accurate-oriented-object | 60.5% |
| an-empirical-study-of-remote-sensing | 76.50% |
| single-stage-rotation-decoupled-detector-for | 77.75% |
| pp-yoloe-r-an-efficient-anchor-free-rotated | 79.42% |
| dota-a-large-scale-dataset-for-object | 52.93% |
| projecting-points-to-axes-oriented-object | 82.26% |
| dense-label-encoding-for-boundary | 77.37% |
| learning-modulated-loss-for-rotated-object | 74.10% |
| mtp-advancing-remote-sensing-foundation-model | 81.66% |
| towards-multi-class-object-detection-in | 68.16% |
| tricubenet-2d-kernel-based-object | 75.26% |
| legnet-lightweight-edge-gaussian-driven | 80.03 |
| lsknet-a-foundation-lightweight-backbone-for | 81.64% |
| polardet-a-fast-more-precise-detector-for | 76.64% |
| oriented-object-detection-in-aerial-images | 75.36% |
| arbitrary-oriented-object-detection-with | 76.17% |
| axis-learning-for-orientated-objects | 65.98% |
| an-empirical-study-of-remote-sensing | 77.38% |
| scrdet-detecting-small-cluttered-and-rotated | 76.81% |
| few-could-be-better-than-all-feature-sampling | 79.59% |
| lwganet-a-lightweight-group-attention | 78.64 |
| anchor-retouching-via-model-interaction-for | 80.37% |
| category-aware-dynamic-label-assignment-with | 82.02% |
| strip-r-cnn-large-strip-convolution-for | 82.28% |
| learning-high-precision-bounding-box-for | 80.63% |