Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks36 from public data
- Shadow optimization from structured deep edge detection88
This paper presents a novel learning-based framework for shadow region recovery from a single image by using a structured CNN learning framework, and proposes and forms shadow/bright measure to model complex interactions among image regions.
- Robust and Resource-Efficient Data-Free Knowledge Distillation by Generative Pseudo Replay63
A Variational Autoencoder with a training objective that is customized to learn the synthetic data representations optimally and optimizes the expected value of the distilled model accuracy while eliminating the large memory overhead incurred by the sample-storing methods.
- Preventing Catastrophic Forgetting and Distribution Mismatch in Knowledge Distillation via Synthetic Data53
A data-free KD framework that maintains a dynamic collection of generated samples over time that can improve the accuracy of the student models obtained via KD when compared with state-of-the-art approaches on the SVHN, Fashion MNIST and CIFAR100 datasets.
- Real-Time Video Stabilization for Unmanned Aerial Vehicles34
A new real-time video stabilization method for Unmanned Aerial Vehicles (UAV) that mainly consists of three steps: keypoints are located based on FAST corner detection and preliminarily matched, motion estimation is performed based on affine transform model and compensation for vibration is conducted based on Spline smoothing.
- 22
- 20
- 19
- 18
- 15
- 12
- 11
- 10
- PDA BASED HUMAN MOTION RECOGNITION SYSTEM10
This paper describes the design and implementation of autonomous real-time motion recognition on a Personal Digital Assistant to test the viability of performing complex video processing on an embedded platform.
- Combination of Local and Global Features for Near-Duplicate Detection9
The results turn out that the proposed method compares favorably against the state-of-the-arts, and has the advantage for handling the case when there are only a few matched keypoints.
- Feature-less Stitching of Cylindrical Tunnel8
This work utilizing known scene geometry along with recorded camera trajectory to create cylindrical images captured in a given environment such as a tunnel where the camera rotates around its center to create an immersive environment for end users.
- Fuzzy rule-based system for dynamic texture and color based background subtraction8
A fuzzy rule-based system that adaptively adjusts the weights of the texture and color features based on the pixel's local properties is proposed, which is based on local binary pattern (LBP) and local color pattern (LCP).
- Fusing Appearance and Spatio-temporal Features for Multiple Camera Tracking7
This work fuse appearance and spatio-temporal features for person re-identification by preprocessing to reduce the number of association candidates and associating objects by using the probabilistic relative distance.
- 7
- 7
- 6
- Hierarchical Audio-Visual Surveillance for Passenger Elevators4
A hierarchical audio-visual surveillance framework for elevators that acts as the front line detector to monitor for events such as urinating, vandalism, and crimes that involved vulnerable targets may not exhibit significant visual cues.
- 3
- Sequential Particle Filter For Multiple Object Tracking2
This paper proposes a jump observation model that includes both visible objects and occluded objects in a sequential particle filter tracking technique that is extended from the sequential estimation method.
- 2
- 2
- Adaptive Background Subtraction with Multiple Feedbacks for Video Surveillance2
A novel framework which integrates both positive and negative feedbacks to control the learning rate is proposed and three strategies for short-term, selective and long-term surveillance have been implemented and tested.
- Condensed Data Expansion Using Model Inversion for Knowledge Distillation1
This work proposes a method that expands condensed datasets using model inversion, a technique for generating synthetic data based on the impressions of a pre-trained model on its training data, which demonstrates significant gains in KD accuracy compared to using condensed datasets alone and outperforms standard model inversion-based KD methods.
- A Novel Human Action Representation via Convolution of Shape-Motion Histograms1
Inspired by the success of convolution-based reverb application in digital signal processing, a novel method to synergistically combine shape and motion histograms via convolution operation is proposed and the resultant feature is more efficient than other hybrid features.
- Robust unmanned aerial vehicle camera self-calibration for surveillance applications1
A new method to calibrate automatically the camera from a few images, using rank 4 factorization to estimate the camera parameters, does not require any known object or known 3D location point in the scene.
- 1
- 1
- –
- –
- –
- –
- Auto-Determination of Camera Parameters for Scenario Simulations in Visual Surveillance Applications–
A novel method for determining geometric parameters of general cameras to generate augmented reality images based on geometry-free characterization of the images using a common codebook and random ferns to find correspondence points is proposed.
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-11. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.