No results for some text prompts

For this image, 'people', 'person', or 'human' do not produce any segmentation results, only use 'kids' as the prompt can get segmentation results.

Reproduce the attached-image case using the prompts “people,” “person,” “human,” and “kids,” and compare the segmentation results. Trace the inference path responsible for text-prompt handling; the issue does not name a file or test. Done means the affected image produces expected segmentation results for the equivalent prompts, with a regression check for the reported behavior.

Written by the indexing model from the issue text.