Abstract
Vision-based human detection is a fundamental task in visual content analysis. It has a wide range of applications, especially for person search and retrieval. To reduce the reliance of detection models on large amount of labeled data, we modify Faster R-CNN to facilitate semi-supervised human detection. Specifically, a Reliability Analysis (RA) module is included as an add-on into our Self-Enhanced R-CNN (SE-RCNN) model. The unlabeled images can be pseudo-annotated reliably under the help of this module. As a result, both labeled and unlabeled data are fed simultaneously for model optimization. The additional supervision, in turn, guides the training of a detection module in our model. The two aspects, extracting precise proposals and generating reliable pseudo annotations, can be mutually reinforced. Unlike previous related works, it is the first attempt to build a single-stage semi-supervised human detection model. In our experiment, we observe that the RA module plays an important role in exploiting unlabeled data and leads to state-of-the-art results of SE-RCNN on multiple benchmarks.
Highlights
As a core component for many promising applications, such as self-driving vehicles, visual surveillance and person reidentification, human detection has attracted considerable attention of both academic and industrial researchers in recent years
The two datasets have been used for evaluation of scene-specific human detection in a number of existing works, since their images are captured in surveillance scenes
When 5% of the training images are annotated, better detection results can be obtained by the semi-supervised methods: ‘Variant SemiBoost’ [14], ‘Temporal Ensembling’ [58] and ‘Self-paced Convolutional Neural Network (CNN)’ [15]
Summary
As a core component for many promising applications, such as self-driving vehicles, visual surveillance and person reidentification, human detection has attracted considerable attention of both academic and industrial researchers in recent years. The main challenges are caused by scale and pose variations, occlusion, and so on [1]. A number of supervised models have been developed for this task [2]–[5]. Considerable progress has been achieved, existing supervised models highly rely on sufficient well-annotated images for training. Safety is critically important for autonomous driving, and lots of efforts have been devoted in annotating training data acquired in highways and urban roads. The detection performance of the resulting models is not satisfactory in complex scenes, while collecting sufficient and strict annotations in the form of bounding boxes in large-scale and various scenarios is very challenging and expensive. It is important to utilize readily available unlabeled data for making detection models to generalize well to unseen data
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
