Abstract:In object detection, the cost of labeling is much high because it needs not only to confirm the categories of multiple objects in an image but also to accurately determine the bounding boxes of each object. Thus, integrating active learning into object detection will raise pretty positive significance. In this paper, we propose a classification committee for active deep object detection method by introducing a discrepancy mechanism of multiple classifiers for samples' selection when training object detectors. The model contains a main detector and a classification committee. The main detector denotes the target object detector trained from a labeled pool composed of the selected informative images. The role of the classification committee is to select the most informative images according to their uncertainty values from the view of classification, which is expected to focus more on the discrepancy and representative of instances. Specifically, they compute the uncertainty for a specified instance within the image by measuring its discrepancy output by the committee pre-trained via the proposed Maximum Classifiers Discrepancy Group Loss (MCDGL). The most informative images are finally determined by selecting the ones with many high-uncertainty instances. Besides, to mitigate the impact of interference instances, we design a Focus on Positive Instances Loss (FPIL) to make the committee the ability to automatically focus on the representative instances as well as precisely encode their discrepancies for the same instance. Experiments are conducted on Pascal VOC and COCO datasets versus some popular object detectors. And results show that our method outperforms the state-of-the-art active learning methods, which verifies the effectiveness of the proposed method.

Latent Dirichlet classification:A new method for object detection

Scene Classification Using Class-Supervised Local-Space-constraint Latent Dirichlet Allocation

Latent Model Ensemble with Auto-localization

Local Attention Sequence Model for Video Object Detection

A Semisupervised Infinite Latent Dirichlet Allocation Model for Target Discrimination in SAR Images with Complex Scenes

Classification Committee for Active Deep Object Detection

Max-Margin Deep Diverse Latent Dirichlet Allocation With Continual Learning

Web Text Classification based on LDA Model

3DLG-Detector: 3D Object Detection via Simultaneous Local-Global Feature Learning

Robust Attribute-Based Visual Recognition Using Discriminative Latent Representation.

Supervised LDA for Image Annotation

Latent Fisher Discriminant Analysis

Partial Membership Latent Dirichlet Allocation

Dataless Text Classification with Descriptive LDA.

A Hybrid Supervised-unsupervised Method on Image Topic Visualization with Convolutional Neural Network and LDA

Latent Topic Visual Language Model for Object Categorization.

LaMI-DETR: Open-Vocabulary Detection with Language Model Instruction

Latent Dirichlet allocation (LDA) and topic modeling: models, applications, a survey

Multi-view Latent Space Learning Based on Local Discriminant Embedding

Feature Extraction By Locality-Based Linear Discriminant Analysis

Subcategory Clustering with Latent Feature Alignment and Filtering for Object Detection