Unverified50% confidenceFactExact time
Policies that rely solely on RGB input will see significantly degraded generalization ability when scenes change, such as when lighting changes or objects are placed at different angles.
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
LeRobot 3D：为SO101机器人构建三维空间感知与操作能力
Related Claims
UnverifiedDice Loss and IoU Loss are more sensitive to the relative proportion of foreground pixels, naturally alleviating class imbalance; Focal Loss forces the model to focus on difficult pixels by dynamically reducing the weight of easily classified samples.68% similarUnverified直接沿用自然图像的数据增强策略（如颜色抖动、随机裁剪）可能破坏雷达数据的物理含义65% similarUnverifiedRGB图像是三维世界在二维平面上的透视投影变换，这个不可逆的降维过程会永久丢失物体的深度、体积和空间拓扑关系64% similarUnverifiedFeature-point methods are more robust to illumination changes and motion blur but discard a large amount of image information, while direct methods can build semi-dense or even dense maps but are sensitive to illumination changes and require good initial values.64% similarUnverified包含明确约束条件、具体交互行为和视觉规范的prompt能够显著减少模型的自由发挥空间，使输出更接近实际需求62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/556849API
curl https://kongchang.com/api/v1/knowledge/claims/556849MCP
get_claim(id=556849)