DC-SAM: In-Context Segment Anything in Images and Videos via Dual Consistency Paper • 2504.12080 • Published Apr 16 • 8