Neural Information Processing Systems http://nips.cc/

Duplicate removal is a critical step to accomplish a reasonable amount of predictions in prevalent proposal-based object detection frameworks.

Duplicate removal is a critical step to accomplish a reasonable amount of predictions in prevalent proposal-based object detection frameworks.

This paper proposes a new Duplicate Removal method based on RNN. Based on each candidate area, informative features are extracted by using appearance feature, position and ranking information in addition to the score. Then, they are treated as series data and are input into the RNN-based model to improve the final accuracy by capturing global information. The number of candidate regions is enormous to the number of objects that are to be left. Therefore, this paper proposes to reduce the box gradually by dividing it into two stages. In the two stages, the RNN model of the same structure was used. In stage I, to remove simple boxes the model is trained by using NMS results as a teaching signal. In stage II, to remove difficult boxes, the model is trained by using the grand-truth boxes. Experiments showed that mAP is increased in the SOTA object detection methods (FPN, Mask R - CNN, PANet with DCN) with the proposed method.

RDF knowledge graphs (KG) are powerful data structures to represent factual statements created from heterogeneous data sources. KG creation is laborious, and demands data management techniques to be executed efficiently. This paper tackles the problem of the automatic generation of KG creation processes declaratively specified; it proposes techniques for planning and transforming heterogeneous data into RDF triples following mapping assertions specified in the RDF Mapping Language (RML). Given a set of mapping assertions, the planner provides an optimized execution plan by partitioning and scheduling the execution of the assertions. First, the planner assesses an optimized number of partitions considering the number of data sources, type of mapping assertions, and the associations between different assertions. After providing a list of partitions and assertions that belong to each partition, the planner determines their execution order. A greedy algorithm is implemented to generate the partitions' bushy tree execution plan. Bushy tree plans are translated into operating system commands that guide the execution of the partitions of the mapping assertions in the order indicated by the bushy tree. The proposed optimization approach is evaluated over state-of-the-art RML-compliant engines and existing benchmarks of data sources and RML triples maps. Our experimental results suggest that the performance of the studied engines can be considerably improved, particularly in a complex setting with numerous triples maps and data sources. As a result, engines that usually time in complex cases out can, if not entirely execute all the assertions, still produce a portion of the KG.

Mainstream object detectors based on the fully convolutional network has achieved impressive performance. While most of them still need a hand-designed non-maximum suppression (NMS) post-processing, which impedes fully end-to-end training. In this paper, we give the analysis of discarding NMS, where the results reveal that a proper label assignment plays a crucial role. To this end, for fully convolutional detectors, we introduce a Prediction-aware One-To-One (POTO) label assignment for classification to enable end-to-end detection, which obtains comparable performance with NMS. Besides, a simple 3D Max Filtering (3DMF) is proposed to utilize the multi-scale features and improve the discriminability of convolutions in the local region. With these techniques, our end-to-end framework achieves competitive performance against many state-of-the-art detectors with NMS on COCO and CrowdHuman datasets. The code is available at https://github.com/Megvii-BaseDetection/DeFCN .

This is Part 2 of the Interview Question series that I recently started. In Part 1, we talked about another important data science interview question pertaining to scaling your ML model. Be sure to check that out! A typical open-ended question that often comes up during interviews (both first and second round) is related to your personal (or side) projects. And trust me when I say this, this question is the best thing that can happen to you during an interview.

Duplicate removal is a critical step to accomplish a reasonable amount of predictions in prevalent proposal-based object detection frameworks. In this work, we design a new two-stage framework to effectively select the appropriate proposal candidate for each object. The first stage suppresses most of easy negative object proposals, while the second stage selects true positives in the reduced proposal set. These two stages share the same network structure, an encoder and a decoder formed as recurrent neural networks (RNN) with global attention and context gate. The encoder scans proposal candidates in a sequential manner to capture the global context information, which is then fed to the decoder to extract optimal proposals.

Duplicate removal is a critical step to accomplish a reasonable amount of predictions in prevalent proposal-based object detection frameworks. Albeit simple and effective, most previous algorithms utilize a greedy process without making sufficient use of properties of input data. In this work, we design a new two-stage framework to effectively select the appropriate proposal candidate for each object. The first stage suppresses most of easy negative object proposals, while the second stage selects true positives in the reduced proposal set. These two stages share the same network structure, i.e., an encoder and a decoder formed as recurrent neural networks (RNN) with global attention and context gate. The encoder scans proposal candidates in a sequential manner to capture the global context information, which is then fed to the decoder to extract optimal proposals. In our extensive experiments, the proposed method outperforms other alternatives by a large margin.

Duplicate removal is a critical step to accomplish a reasonable amount of predictions inprevalent proposal-based object detection frameworks. Albeit simple and effective, most previous algorithms utilize a greedy process without making sufficient useof properties of input data. In this work, we design a new two-stage framework to effectively select the appropriate proposal candidate for each object. Thefirst stage suppresses most of easy negative object proposals, while the second stage selects true positives in the reduced proposal set. These two stages share the same network structure, i.e., an encoder and a decoder formed as recurrent neuralnetworks (RNN) with global attention and context gate. The encoder scans proposal candidates in a sequential manner to capture the global context information, whichis then fed to the decoder to extract optimal proposals. In our extensive experiments, the proposed method outperforms other alternatives by a large margin.