in illumination, occlusions, pose variations, and even the absence of visual query. In this thesis,

Cross-modal person re-identification (Re-ID) is a crucial component of a modern video surveillance

The first contribution proposes to jointly model the multi-modal latent space, where corresponding

Unified feature learning effectively utilizes textual data as a super-annotation signal for visual

has further solidified the idea of a unified backbone model. In the final contribution, we propose

backbone to implicitly align the shared semantic concepts from the start of the learning network.

a vision transformer architecture design with the aim of an effective intra-modal and cross-modal

late-fusion models depends on the quality of the feature extraction backbones for each modality.

system and security infrastructure. The task of matching people across multiple nonoverlapping

key challenge is to align cross-modality feature representations according to the fine-grained