Hakan Bilen

Dynamic Image Networks for Action Recognition

We introduce the concept of dynamic image, a novel compact representation of videos useful for video analysis especially when convolutional neural networks (CNNs) are used. The dynamic image is based on the rank pooling concept and is obtained through the parameters of a ranking machine that encodes the temporal evolution of the frames of the video. Dynamic images are obtained by directly applying rank pooling on the raw image pixels of a video producing a single RGB image per video. This idea is simple but powerful as it enables the use of existing CNN models directly on video data with fine-tuning. We present an efficient and effective approximate rank pooling operator, speeding it up orders of magnitude compared to rank pooling. Our new approximate rank pooling CNN layer allows us to generalize dynamic images to dynamic feature maps and we demonstrate the power of our new representations on standard benchmarks in action recognition achieving state-of-the-art performance.

Weakly Supervised Deep Detection Networks

Weakly supervised learning of object detection is an important problem in image understanding that still does not have a satisfactory solution. In this paper, we address this problem by exploiting the power of deep convolutional neural networks pre-trained on large-scale image-level classification tasks. We propose a weakly supervised deep detection architecture that modifies one such network to operate at the level of image regions, performing simultaneously region selection and classification. Trained as an image classifier, the architecture implicitly learns object detectors that are better than alternative weakly supervised detection systems on the PASCAL VOC data. The model, which is a simple and elegant end-to-end architecture, outperforms standard data augmentation and fine-tuning techniques for the task of image-level classification as well.

Weakly supervised object detection with convex clustering

Weakly supervised object detection, is a challenging task, where the training procedure involves learning at the same time both, the model appearance and the object location in each image. The classical approach to solve this problem is to consider the location of the object of interest in each image as a latent variable and minimize the loss generated by such latent variable during learning. However, as learning appearance and localization are two interconnected tasks, the optimization is not convex and the procedure can easily get stuck in a poor local minimum, i.e. the algorithm “misses” the object in some images. In this paper, we help the optimization to get close to the global minimum by enforcing a “soft” similarity between each possible location in the image and a reduced set of “exemplars”, or clusters, learned with a convex formulation in the training images. The help is effective because it comes from a different and smooth source of information that is not directly connected with the main task. Results show that our method improves a strong baseline based on convolutional neural network features by more than 4 points without any additional features or extra computation at testing time but only adding a small increment of the training time due to the convex clustering.

Self-Supervised Video Representation Learning with Odd-One-Out Networks

We propose a new self-supervised CNN pre-training technique based on a novel auxiliary task called odd-one-out learning. In this task, the machine is asked to identify the unrelated or odd element from a set of otherwise related elements. We apply this technique to self-supervised video representation learning where we sample subsequences from videos and ask the network to learn to predict the odd video subsequence. The odd video subsequence is sampled such that it has wrong temporal order of frames while the even ones have the correct temporal order. Therefore, to generate a odd-one-out question no manual annotation is required. Our learning machine is implemented as multi-stream convolutional neural network, which is learned end-to-end. Using odd-one-out networks, we learn temporal representations for videos that generalizes to other related tasks such as action recognition. On action classification, our method obtains 60.3% on the UCF101 dataset using only UCF101 data for training which is approximately 10% better than current state-of-the-art self-supervised learning methods. Similarly, on HMDB51 dataset we outperform self-supervised state-of-the art methods by 12.7% on action classification task.

Object and Action Classification with Latent Variables

In this paper we propose a generic framework to incorporate unobserved auxiliary information for classifying objects and actions. This framework allows us to explicitly account for localisation and alignment of representations for generic object and action classes as latent variables. We approach this problem in the discriminative setting as learning a max-margin classifier that infers the class label along with the latent variables. Through this paper we make the following contributions a) We provide a method for incorporating latent variables into object and action classification b) We specifically account for the presence of an explicit class related subregion which can include foreground and/or background. c) We explore a way to learn a better classifier by iterative expansion of the latent parameter space. We demonstrate the performance of our approach by rigorous experimental evaluation on a number of standard object and action recognition datasets.

Micromanipulation Using a Microassembly Workstation with Vision and Force Sensing

This paper reports our ongoing work on a microassembly workstation developed for efficient and robust 3D automated assembly of microobjects. The workstation consists of multiple view imaging system, two 3-DOF high precision micromanipulators, and a 3-DOF positioning stage with high resolution rotation control, force sensing probe and gripper, and the control software system. A hybrid control scheme using both vision and force sensory information is proposed for precise and dexterous manipulation of microobjects. A micromanipulation experiment that aims to locate the microspheres to the predefined configuration by using an integrated vision and force control scheme is successfully demonstrated to show the validity of the proposed methods.

Object and Action Classification with Latent Window Parameters

In this paper we propose a generic framework to incorporate unobserved auxiliary information for classifying objects and actions. This framework allows us to automatically select a bounding box and its quadrants from which best to extract features. These spatial subdivisions are learnt as latent variables. The paper is an extended version of our earlier work Bilen et al. (Proceedings of The British Machine Vision Conference, 2011), complemented with additional ideas, experiments and analysis. We approach the classification problem in a discriminative setting, as learning a max-margin classifier that infers the class label along with the latent variables. Through this paper we make the following contributions: (a) we provide a method for incorporating latent variables into object and action classification; (b) these variables determine the relative focus on foreground versus background information that is taken account of; (c) we design an objective function to more effectively learn in unbalanced data sets; (d) we learn a better classifier by iterative expansion of the latent parameter space. We demonstrate the performance of our approach through experimental evaluation on a number of standard object and action recognition data sets.

Developing robust vision modules for microsystems applications

In this work, several robust vision modules are developed and implemented for fully automated micromanipulation. These are autofocusing, object and end-effector detection, real-time tracking and optical system calibration modules. An image based visual servoing architecture and a path planning algorithm are also proposed based on the developed vision modules. Experimental results are provided to assess the performance of the proposed visual servoing approach in positioning and trajectory tracking tasks. Proposed path planning algorithm in conjunction with visual servoing imply successful micromanipulation tasks.

Action Recognition with Dynamic Image Networks

We introduce the concept of dynamic image, a novel compact representation of videos useful for video analysis, particularly in combination with convolutional neural networks (CNNs). A dynamic image encodes temporal data such as RGB or optical flow videos by using the concept of ‘rank pooling’. The idea is to learn a ranking machine that captures the temporal evolution of the data and to use the parameters of the latter as a representation. We call the resulting representation dynamic image because it summarizes the video dynamics in addition to appearance. This powerful idea allows to convert any video to an image so that existing CNN models pre-trained with still images can be immediately extended to videos. We also present an efficient approximate rank pooling operator that runs two orders of magnitude faster than the standard ones with any loss in ranking performance and can be formulated as a CNN layer. To demonstrate the power of the representation, we introduce a novel four stream CNN architecture which can learn from RGB and optical flow frames as well as from their dynamic image representations. We show that the proposed network achieves state-of-the-art performance, 95.5 and 72.5 percent accuracy, in the UCF101 and HMDB51, respectively.

Action recognition: A region based approach

We address the problem of recognizing actions in reallife videos. Space-time interest point-based approaches have been widely prevalent towards solving this problem. In contrast, more spatially extended features such as regions have not been so popular. The reason is, any local region based approach requires the motion flow information for a specific region to be collated temporally. This is challenging as the local regions are deformable and not well delineated from the surroundings. In this paper we address this issue by using robust tracking of regions and we show that it is possible to obtain region descriptors for classification of actions. This paper lays the groundwork for further investigation into region based approaches. Through this paper we make the following contributions a) We advocate identification of salient regions based on motion segmentation b) We adopt a state-of-the art tracker for robust tracking of the identified regions rather than using isolated space-time blocks c) We propose optical flow based region descriptors to encode the extracted trajectories in piece-wise blocks. We demonstrate the performance of our system on real-world data sets.