@orhir.bsky.social
4/ We hope EdgeCape inspires new ideas in category-agnostic pose estimation research. 🙌
Feel free to reach out if you have questions or are interested in collaborations! I'd love to hear your thoughts.
Feel free to reach out if you have questions or are interested in collaborations! I'd love to hear your thoughts.
November 26, 2024 at 9:01 AM
4/ We hope EdgeCape inspires new ideas in category-agnostic pose estimation research. 🙌
Feel free to reach out if you have questions or are interested in collaborations! I'd love to hear your thoughts.
Feel free to reach out if you have questions or are interested in collaborations! I'd love to hear your thoughts.
3/ 📊 Results:
EdgeCape outperforms state-of-the-art methods on the MP-100 benchmark, achieving:
✅ New SOTA in 1-shot settings
✅ Superior performance among similar-sized methods in 5-shot settings
We also shine in cross-category and occlusion scenarios! 💪
EdgeCape outperforms state-of-the-art methods on the MP-100 benchmark, achieving:
✅ New SOTA in 1-shot settings
✅ Superior performance among similar-sized methods in 5-shot settings
We also shine in cross-category and occlusion scenarios! 💪
November 26, 2024 at 9:01 AM
3/ 📊 Results:
EdgeCape outperforms state-of-the-art methods on the MP-100 benchmark, achieving:
✅ New SOTA in 1-shot settings
✅ Superior performance among similar-sized methods in 5-shot settings
We also shine in cross-category and occlusion scenarios! 💪
EdgeCape outperforms state-of-the-art methods on the MP-100 benchmark, achieving:
✅ New SOTA in 1-shot settings
✅ Superior performance among similar-sized methods in 5-shot settings
We also shine in cross-category and occlusion scenarios! 💪
2/ What's new?
Most methods use either isolated keypoints or fixed, unweighted pose graphs. This limits their ability to fully leverage structural priors.
💡 Enter EdgeCape:
✅ Predicts weighted graphs
✅ Enhanced graph-structure utilization
✅ Better feature extraction
Most methods use either isolated keypoints or fixed, unweighted pose graphs. This limits their ability to fully leverage structural priors.
💡 Enter EdgeCape:
✅ Predicts weighted graphs
✅ Enhanced graph-structure utilization
✅ Better feature extraction
November 26, 2024 at 9:01 AM
2/ What's new?
Most methods use either isolated keypoints or fixed, unweighted pose graphs. This limits their ability to fully leverage structural priors.
💡 Enter EdgeCape:
✅ Predicts weighted graphs
✅ Enhanced graph-structure utilization
✅ Better feature extraction
Most methods use either isolated keypoints or fixed, unweighted pose graphs. This limits their ability to fully leverage structural priors.
💡 Enter EdgeCape:
✅ Predicts weighted graphs
✅ Enhanced graph-structure utilization
✅ Better feature extraction
🧵 Let's dive in:
1/ CAPE aims to generalize pose estimation to any object category using just 1 or a few annotated support images. It's the key to unlocking pose estimation for the long tail of object categories! 🚀
However, current methods struggle with occlusions & symmetry.
1/ CAPE aims to generalize pose estimation to any object category using just 1 or a few annotated support images. It's the key to unlocking pose estimation for the long tail of object categories! 🚀
However, current methods struggle with occlusions & symmetry.
November 26, 2024 at 9:01 AM
🧵 Let's dive in:
1/ CAPE aims to generalize pose estimation to any object category using just 1 or a few annotated support images. It's the key to unlocking pose estimation for the long tail of object categories! 🚀
However, current methods struggle with occlusions & symmetry.
1/ CAPE aims to generalize pose estimation to any object category using just 1 or a few annotated support images. It's the key to unlocking pose estimation for the long tail of object categories! 🚀
However, current methods struggle with occlusions & symmetry.