Learning Aerial Image Segmentation from Online Maps

This study deals with semantic segmentation of high-resolution (aerial) images where a semantic class label is assigned to each pixel via supervised classification as a basis for automatic map generation. Recently, deep convolutional neural networks (CNNs) have shown impressive performance and have quickly become the de-facto standard for semantic segmentation, with the added benefit that task-specific feature design is no longer necessary. However, a major downside of deep learning methods is that they are extremely data-hungry, thus aggravating the perennial bottleneck of supervised classification, to obtain enough annotated training data. On the other hand, it has been observed that they are rather robust against noise in the training labels. This opens up the intriguing possibility to avoid annotating huge amounts of training data, and instead train the classifier from existing legacy data or crowd-sourced maps which can exhibit high levels of noise. The question addressed in this paper is: can training with large-scale, publicly available labels replace a substantial part of the manual labeling effort and still achieve sufficient performance? Such data will inevitably contain a significant portion of errors, but in return virtually unlimited quantities of it are available in larger parts of the world. We adapt a state-of-the-art CNN architecture for semantic segmentation of buildings and roads in aerial images, and compare its performance when using different training data sets, ranging from manually labeled, pixel-accurate ground truth of the same city to automatic training data derived from OpenStreetMap data from distant locations. We report our results that indicate that satisfying performance can be obtained with significantly less manual annotation effort, by exploiting noisy large-scale training data.

Neural Scene Decomposition for Multi-Person Motion Capture

Victor Constantin, Pascal Fua, Isinsu Katircioglu, Helge Jochen Rhodin, Mathieu Salzmann

Learning general image representations has proven key to the success of many computer vision tasks. For example, many approaches to image understanding problems rely on deep networks that were initially trained on ImageNet, mostly because the learned features are a valuable starting point to learn from limited labeled data. However, when it comes to 3D motion capture of multiple people, these features are only of limited use. In this paper, we therefore propose an approach to learning features that are useful for this purpose. To this end, we introduce a self-supervised approach to learning what we call a neural scene decomposition (NSD) that can be exploited for 3D pose estimation. NSD comprises three layers of abstraction to represent human subjects: spatial layout in terms of bounding-boxes and relative depth; a 2D shape representation in terms of an instance segmentation mask; and subject-specific appearance and 3D pose information. By exploiting self-supervision coming from multiview data, our NSD model can be trained end-to-end without any 2D or 3D supervision. In contrast to previous approaches, it works for multiple persons and full-frame images. Because it encodes 3D geometry, NSD can then be effectively leveraged to train a 3D pose estimation network from small amounts of annotated data. Our code and newly introduced boxing dataset is available at github.com and cvlab.epfl.ch.

IEEE2019

Learning to Detect Objects with Minimal Supervision

Karim Ali

Many classes of objects can now be successfully detected with statistical machine learning techniques. Faces, cars and pedestrians, have all been detected with low error rates by learning their appearance in a highly generic manner from extensive training sets. These recent advances have enabled the use of reliable object detection components in real systems, such as automatic face focusing functions on digital cameras. One key drawback of these methods, and the issue addressed here, is the prohibitive requirement that training sets contain thousands of manually annotated examples. We present three methods which make headway toward reducing labeling requirements and in turn, toward a tractable solution to the general detection problem. First, we propose a new learning strategy for object detection. The proposed scheme forgoes the need to train a collection of detectors dedicated to homogeneous families of poses, and instead learns a single classifier that has the inherent ability to deform based on the signal of interest. We train a detector with a standard AdaBoost procedure by using combinations of pose-indexed features and pose estimators. This allows the learning process to select and combine various estimates of the pose with features able to compensate for variations in pose without the need to label data for training or explore the pose space in testing. We validate our framework on three types of data: hand video sequences, aerial images of cars, as well as face images. We compare our method to a standard Boosting framework, with access to the same ground truth, and show a reduction in the false alarm rate of up to an order of magnitude. Where possible, we compare our method to the state-of-the art, which requires pose annotations of the training data, and demonstrate comparable performance. Second, we propose a new learning method which exploits temporal consistency to successfully learn a complex appearance model from a sparsely labeled training video. Our approach consists in iteratively improving an appearance-based model built with a Boosting procedure, and the reconstruction of trajectories corresponding to the motion of multiple targets. We demonstrate the efficiency of our procedure by learning a pedestrian detector from videos and a cell detector from microscopy image sequences. In both cases, our method is demonstrated to reduce the labeling requirement by one to two orders of magnitude. We show that in some instances, our method trained with sparse labels on a video sequence is able to outperform a standard learning procedure trained with the fully labeled sequence. Third, we propose a new active learning procedure which exploits the spatial structure of image data and queries entire scenes or frames of a video rather than individual examples. We extend the Query by Committee approach allowing it to characterize the most informative scenes that are to be selected for labeling. We show that an aggressive procedure which exhibits zero tolerance to target localization error performs as well as more sophisticated strategies taking into account the trade-off between missed detections and localization error. Finally, we combine this method with our two proposed approaches above and demonstrate that the resulting algorithm can properly perform car detection from a small set of annotated image as well as pedestrian detection from a handful of labeled video frames.

EPFL2012

From Human-Designed Convolutional Neural Networks Towards Robust Neural Architecture Search

Kaicheng Yu

Artificial intelligence has been an ultimate design goal since the inception of computers decades ago. Among the many attempts towards general artificial intelligence, modern machine learning successfully tackles many complex problems thanks to the progress in deep learning, and in particular in convolutional neural networks (CNN). To design a CNN for a specific task, one common approach consists of adapting the heuristics from the pre-deep-learning era to the CNN domain. In the first part of this thesis, we introduce two methods that follow this approach: i) We build a covariance descriptor, i.e., a local descriptor that is suitable for texture recognition, to replace the first-order fully connected layers in an ordinary CNN, showing that such a descriptor yields state-of-the-art performance on many fine-grained image classification tasks with orders of magnitude fewer feature dimensions; ii) we develop a light-weight recurrent U-Net for image semantic segmentation, inspired by the biological eye saccadic movements, that yields real-time predictions on devices with limited computational resources. As most methods pre-dating automatic machine learning (AutoML), the two above-mentioned CNNs were human-designed. In the past few years, however, neural architecture search~(NAS), which aims to facilitate the design of deep networks for new tasks, has drawn an increasing attention. In this context, the weight-sharing approach, which consists of utilizing a super-net to encompass all possible architectures within a search space, has become a de facto standard in NAS because it enables the search to be done on commodity hardware. In the second part of this thesis, we then provide an in-depth study of recent weight-sharing NAS algorithms. First, we discover a phenomenon in the weight-sharing NAS training pipeline, which we dub multi-model forgetting, that negatively impacts the super-net quality, and propose a statistically motivated approach to address it. Subsequently, we find that (i) on average, many popular NAS algorithms perform similarly to a random architecture sampling policy; (ii) the widely-adopted weight sharing strategy degrades the ranking of the NAS candidates to the point of not reflecting their true performance, thus reducing the effectiveness of the search process. We then further decouple weight sharing from the NAS sampling policy, and isolate 14 factors that play a key role in the success of super-net training. Finally, to improve the super-net quality, we propose a regularization term that aims to maximize the correlation between the performance rankings of the super-net and of the stand-alone architectures using a small set of landmark architectures.