Study on an infrared multi-target detection method based on the pseudo-two-stage model

Zhou, Tong; Yu, Zhentao; Cao, Yu; Bai, Hongyang; Su, Yan

doi:10.1016/j.infrared.2021.103883

Cited by 9 publications

(10 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More importantly, SMG-Y with 1/4 training data also achieves 74.5% mAP, surpassing all visual detectors and most infrared detectors trained on full FLIR. The bicycle accuracy in SMG-Y-1/4 is 62.5% AP, which is on par with that of Pseudo-two-stage [14]. Note that the training dataset of SMG-C-1/4 only contains 928 bicycle instances, while Pseudo-twostage [14] is trained with 3986 examples for bicycle detection.…”

Section: Comparison With State-of-the-artsmentioning

confidence: 85%

“…The bicycle accuracy in SMG-Y-1/4 is 62.5% AP, which is on par with that of Pseudo-two-stage [14]. Note that the training dataset of SMG-C-1/4 only contains 928 bicycle instances, while Pseudo-twostage [14] is trained with 3986 examples for bicycle detection.…”

Section: Comparison With State-of-the-artsmentioning

confidence: 85%

“…For infrared detection, the lack of large-scale labeled thermal images hinders the power of detectors based on CNN. Researchers cope with this problem from two aspects: one is finetuning a pre-trained model [14,15], the other is introducing corresponding RGB images as supplements [1,16]. The first strategy hardly makes full use of the information from the RGB domain, and the sophisticated structures in the second method are difficult to be performed.…”

Section: Object Detectionmentioning

confidence: 99%

“…The visual detectors such as SSD [9], YOLOv3 [18], Faster-RCNN [8], CenterNet [19], and RefineNet [39] are designed for RGB object detection and finetuned on the training set of FLIR. The infrared detectors, including MMTOD-CG [1], MMTOD-UNIT [1], Effi-YOLOv3 [40] and Pesudo-two-stage [14], are applied for thermal images directly.…”

Section: Comparison With State-of-the-artsmentioning

confidence: 99%

“…To extract infrared object features better, most of the infrared detectors improve existing detection frameworks by introducing some extra enhanced modules such as feature fusion and background suppression. For example, Zhou et al [14] apply a dual cascade regression mechanism to fuse high-level and low-level features. Miao et al [15] design an auxiliary foreground prediction loss to reduce background interference.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Borrow from Source Models: Efficient Infrared Object Detection with Limited Examples

Chen

Liu

et al. 2022

Applied Sciences

View full text Add to dashboard Cite

Recent deep models trained on large-scale RGB datasets lead to considerable achievements in visual detection tasks. However, the training examples are often limited for an infrared detection task, which may deteriorate the performance of deep detectors. In this paper, we propose a transfer approach, Source Model Guidance (SMG), where we leverage a high-capacity RGB detection model as the guidance to supervise the training process of an infrared detection network. In SMG, the foreground soft label generated from the RGB model is introduced as source knowledge to provide guidance for cross-domain transfer. Additionally, we design a Background Suppression Module in the infrared network to receive the knowledge and enhance the foreground features. SMG is easily plugged into any modern detection framework, and we show two explicit instantiations of it, SMG-C and SMG-Y, based on CenterNet and YOLOv3, respectively. Extensive experiments on different benchmarks show that both SMG-C and SMG-Y achieve remarkable performance even if the training set is scarce. Compared to advanced detectors on public FLIR, SMG-Y with 77.0% mAP outperforms others in accuracy, and SMG-C achieves real-time detection at a speed of 107 FPS. More importantly, SMG-Y trained on a quarter of the thermal dataset obtains 74.5% mAP, surpassing most state-of-the-art detectors with full FLIR as training data.

show abstract

Section: Comparison With State-of-the-artsmentioning

confidence: 85%

Section: Comparison With State-of-the-artsmentioning

confidence: 85%

Section: Object Detectionmentioning

confidence: 99%

Section: Comparison With State-of-the-artsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Borrow from Source Models: Efficient Infrared Object Detection with Limited Examples

Chen

Liu

et al. 2022

Applied Sciences

View full text Add to dashboard Cite

show abstract

Improved deep learning based multi‐index comprehensive evaluation system for the research on accuracy and real‐time performance of coal gangue particles recognition

Zhang,

Yang,

Zeng

2023

Energy Science & Engineering

View full text Add to dashboard Cite

The problems of insufficient recognition accuracy, poor real‐time performance and lack of consideration of actual working conditions in the process of intelligent construction of coal mines make this technology still in the research stage and not applied in practical engineering. The purpose of this paper is to establish an accurate and real‐time recognition model, which can quickly distinguish the vibration acceleration signals of coal and gangue under the influence of external factors such as impact position, velocity, and direction by using the different physical properties of coal and gangue particles. Therefore, the accuracy and real‐time of coal gangue recognition model established by different convolutional neural networks (CNN) structures and different position signal input are studied. First, to meet the real‐time requirements, an original CNN recognition model composed of single convolution layer and single pooling layer is established, and the data collected by seven sensors are input in the form of two‐dimensional matrix. However, the stability of the training and test results is insufficient. To solve this problem, once improved CNN (OI‐CNN) recognition model with multiconvolution layers and multipooling layers is built by deepening the network. The experimental results show that the stability and accuracy are improved, but the real‐time performance is poor. Furthermore, through parameter adjustment, the OI‐CNN is changed to the twice improved CNN (TI‐CNN), and the sensor data at different positions are input in the form of one‐dimensional vectors. The results show that the accuracy and real‐time performance of the TI‐CNN coal gangue recognition model are further improved. Finally, according to the research purpose of this paper, the weights of CNN indexes are given, and a multi‐index comprehensive evaluation system (MICES) is established. With the original CNN recognition model as the control, the OI‐CNN recognition model and the TI‐CNN recognition model at different positions are quantitatively compared to obtain the comprehensive evaluation scores of each model. The results show that the MICES of the coal gangue recognition model established based on the TI‐CNN structure and the data input of a single position sensor is the highest, while the sensor position has little effect on the recognition results.

show abstract

Lost in Compression: the Impact of Lossy Image Compression on Variable Size Object Detection within Infrared Imagery

Bhowmik

Barker

Gaus

et al. 2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

View full text Add to dashboard Cite

Lossy image compression strategies allow for more efficient storage and transmission of data by encoding data to a reduced form. This is essential enable training with larger datasets on less storage-equipped environments. However, such compression can cause severe decline in performance of deep Convolution Neural Network (CNN) architectures even when mild compression is applied and the resulting compressed imagery is visually identical. In this work, we apply the lossy JPEG compression method with six discrete levels of increasing compression {95, 75, 50, 15, 10, 5} to infrared band (thermal) imagery. Our study quantitatively evaluates the affect that increasing levels of lossy compression has upon the performance of characteristically diverse object detection architectures (Cascade-RCNN, FSAF and Deformable DETR) with respect to varying sizes of objects present in the dataset. When training and evaluating on uncompressed data as a baseline, we achieve maximal mean Average Precision (mAP) of 0.823 with Cascade R-CNN across the FLIR dataset, outperforming prior work. The impact of the lossy compression is more extreme at higher compression levels (15, 10, 5) across all three CNN architectures. However, re-training models on lossy compressed imagery notably ameliorated performances for all three CNN models with an average increment of ∼ 76% (at higher compression level 5). Additionally, we demonstrate the relative sensitivity of differing object areas {tiny, small, medium, large} with respect to the compression level. We show that tiny and small objects are more sensitive to compression than medium and large objects. Overall, Cascade R-CNN attains the maximal mAP across most of the object area categories.

show abstract

Study on an infrared multi-target detection method based on the pseudo-two-stage model

Cited by 9 publications

References 14 publications

Borrow from Source Models: Efficient Infrared Object Detection with Limited Examples

Borrow from Source Models: Efficient Infrared Object Detection with Limited Examples

Improved deep learning based multi‐index comprehensive evaluation system for the research on accuracy and real‐time performance of coal gangue particles recognition

Lost in Compression: the Impact of Lossy Image Compression on Variable Size Object Detection within Infrared Imagery

Contact Info

Product

Resources

About