Attribute-based Person Retrieval and Search in Video Sequences

Visual appearance based person retrieval in unconstrained environment videos

“…It can be observed from Fig. 10b that the algorithm also achieves similar performance in terms of detection with IoU ≥ 0.4 when compared to state-of-the-art approach of [16]. The use of tracking in [16] helped to achieve better performance for very easy, easy and medium level while the proposed approach outperforms [16] for the hard level.…”

“…It can be observed from Fig. 10a that the proposed approach outperforms all previous approaches [14][15][16] for all difficulty levels in terms of average IoU. The performance increases by 14.69%, 1.32%, 13.12% and 10.35% for difficulty levels; very easy, easy, medium, and hard respectively over the state-of-the-art methods [14][15][16].…”

“…Deep learning based methodologies are widely used for semantic description based person retrieval due to its efficient learning. Convolutional Neural Network (CNN) based algorithms are reported for person attribute recognition [10][11][12] as well as for person retrieval [5,[13][14][15][16]. Semantic Retrieval Convolution Neural Network (SRCNN) [10] retrieves the soft biometrics with recognition rate of 20.1% and 46.4% at rank-1 for one-shot and multi-shot identification respectively on SoBiR [8] dataset.…”

“…The retrieved subject with the smallest loss is the target. Schumann et al [16] use Single-Shot Multibox Detector (SSD) for person detection and refine it using degree of motion in the SSD detected area by background modeling. Such detection methods fail to locate non-moving persons.…”

“…The paper uses one more model for pose estimation while extracting the patch for color and texture. In the AVSS challenge II, [14] achieved state-of-the-art per formance with an average IoU of 0.51 and [16] had 75.9% of frames with an IoU greater than 0.4. The use of tracking in the video sequences provides superior detection [16] .…”