RT-DETR important words 31-150 (3-part)

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/119

encourage image

There's no tags or description

Looks like no tags are added yet.

120 Terms

1
New cards

accommodate /əˈkɑməˌdeɪt/

【動詞】対応する、収める | 例文:Furthermore, RT-DETR supports flexible speed tuning to accommodate various real-time scenarios without retraining, thanks to the multi-layer decoder architecture of DETR.(Introduction)

2
New cards

achieves /əˈtʃiːvz/

【動詞】達成する(3人称単数現在)(基本形:achieve) | 例文:Our RT-DETR-R50 / R101 achieves 53.1% / 54.3% AP on COCO and 108 / 74 FPS on T4 GPU, outperforming previously advanced YOLOs in both speed and accuracy.(Abstract)

3
New cards

adjusting /əˈdʒʌstɪŋ/

【動詞】調整する(〜ing形)(基本形:adjust) | 例文:In addition, RT-DETR supports flexible speed tuning by adjusting the number of decoder layers to adapt to various scenarios without retraining.(Abstract)

4
New cards

adopt /əˈdɑpt/

【動詞】採用する | 例文:However, we observe that the current query selection directly adopt classification scores for selection, ignoring the fact that the detector are required to simultaneously model the category and location of objects, both of which determine the quality of the features.(Introduction)

5
New cards

analyze /ˈænəˌlaɪz/

【動詞】分析する、解析する | 例文:We quantitatively analyze the impact of NMS on the speed and accuracy of YOLO detectors, and establish an end-to-end speed benchmark to test the end-to-end inference speed of real-time detectors.(Introduction)

6
New cards

compare /kəmˈpɛr/

【動詞】比較する | 例文:We compare the scaled RT-DETRs with the lighter (S and M ) YOLO detectors in Appendix, which outperform all S and M models in both speed and accuracy.(Method)

7
New cards

concatenate /kənˈkætəˌneɪt/

【動詞】連結する(チャンネル方向に結合する) | 例文:The multi-scale features share the encoder for intra-scale feature interaction and then concatenate as output.(Method)

8
New cards

conclude /kənˈklud/

【動詞】結論づける、結論を出す | 例文:From the results, we can conclude that the execution time of the EfficientNMS kernel increases as the confidence threshold decreases or the IoU threshold increases.(End-to-end Speed of Detectors)

9
New cards

conduct /kənˈdʌkt/

【動詞】実施する、行う | 例文:To achieve the above goal, we rethink DETRs and conduct detailed analysis of key components to reduce unnecessary computational redundancy and further improve accuracy.(Introduction)

10
New cards

decouples /ˌdiˈkʌpəlz/

【動詞】分離する、切り離す(基本形:decouple) | 例文:Variant D decouples intra-scale interaction and cross-scale fusion by utilizing the single-scale Transformer encoder for the former and a PANet-style [22] structure for the latter.(Method)

11
New cards

enhances /ɛnˈhænsɪz/

【動詞】向上させる、強化する(基本形:enhance) | 例文:Lite DETR [18] enhances the efficiency of encoder by reducing the update frequency of low-level features in an interleaved way.(Related Work)

12
New cards

establish /ɪˈstæblɪʃ/

【動詞】確立する、示す | 例文:We quantitatively analyze the impact of NMS on the speed and accuracy of YOLO detectors, and establish an end-to-end speed benchmark to test the end-to-end inference speed of real-time detectors.(Introduction)

13
New cards

evaluate /ɪˈvæljəˌweɪt/

【動詞】評価する、検証する | 例文:Furthermore, we use YOLOv8 to evaluate the accuracy on the COCO val2017 and test the execution time of the NMS operation under different hyperparameters.(End-to-end Speed of Detectors)

14
New cards

exceeds /ɪkˈsiːdz/

【動詞】超える、上回る(基本形:exceed) | 例文:Specifically, the boxes with scores below the confidence threshold are directly filtered out, and whenever the IoU of any two boxes exceeds the IoU threshold, the box with the lower score will be discarded.(End-to-end Speed of Detectors)

15
New cards

flatten /ˈflætən/

【動詞】平坦化する(1次元に展開する) | 例文:We formulate the calculation of the hybrid encoder as: Q = K = V = Flatten(S5), F5 = Reshape(AIFI(Q, K, V)), O = CCFF({S3, S4, F5}), where Reshape represents restoring the shape of the flattened feature to the same shape as S5.(Method)

16
New cards

generate /ˈdʒɛnəˌreɪt/

【動詞】生成する、生み出す | 例文:Finally, the decoder with auxiliary prediction heads iteratively optimizes object queries to generate categories and boxes.(Method)

17
New cards

improves /ɪmˈpruːvz/

【動詞】改善する(3人称単数現在)(基本形:improve) | 例文:The introduction of multi-scale features accelerates training convergence and improves performance [45].(Method)

18
New cards

increases /ɪnˈkriːsɪz/

【動詞】増える、増加する(基本形:increase) | 例文:As the confidence threshold increases, more prediction boxes are filtered out, and the number of remaining boxes that need to calculate IoU decreases, thus reducing the execution time of NMS.(End-to-end Speed of Detectors)

19
New cards

introduces /ˌɪntrəˈdusɪz/

【動詞】導入する、提示する(基本形:introduce) | 例文:However, these detectors typically require Non-Maximum Suppression (NMS) for post-processing, which not only slows down the inference speed but also introduces hyperparameters that cause instability in both the speed and accuracy.(Introduction)

20
New cards

minimize /ˈmɪnɪmaɪz/

【動詞】最小化する | 例文:To minimize the uncertainty of the queries, we integrate the uncertainty into the loss function for the gradient-based optimization in Eq. (3).(Method)

21
New cards

observe /əbˈzɝv/

【動詞】観察する、認める | 例文:However, we observe that the speed and accuracy of YOLOs are negatively affected by the NMS.(Abstract)

22
New cards

overcome /ˌoʊvərˈkʌm/

【動詞】克服する、乗り越える | 例文:To overcome this bottleneck, we first analyze the computational redundancy present in the multi-scale Transformer encoder.(Method)

23
New cards

propose /prəˈpoʊz/

【動詞】提案する | 例文:In this paper, we propose the Real-Time DEtection TRansformer (RT-DETR), the first real-time end-to-end object detector to our best knowledge that addresses the above dilemma.(Abstract)

24
New cards

reduces /rɪˈduːsɪz/

【動詞】減らす、削減する(基本形:reduce) | 例文:However, although the deformable attention reduces the computational cost, the sharply increased sequence length still causes the encoder to become the computational bottleneck.(Method)

25
New cards

refers /rɪˈfɜːrz/

【動詞】指す、言及する(「〜を指す」の現在形)(基本形:refer) | 例文:Note that the NMS operation we adopt refers to the TensorRT efficientNMSPlugin, which involves multiple kernels, including EfficientNMSFilter, RadixSort, EfficientNMS, etc., and we only report the execution time of the EfficientNMS kernel.(End-to-end Speed of Detectors)

26
New cards

replace /rɪˈpleɪs/

【動詞】置き換える、置換する | 例文:And for the latter, previous works [42, 44, 45] show that the hard-to-optimize object queries hinder the performance of DETRs and propose the query selection schemes to replace the vanilla learnable embeddings with encoder features.(Introduction)

27
New cards

represents /ˌrɛprɪˈzɛnts/

【動詞】表す、示す(基本形:represent) | 例文:SSE represents the single-scale Transformer encoder, MSE represents the multi-scale Transformer encoder, and CSF represents cross-scale fusion.(End-to-end Speed of Detectors)

28
New cards

require /rɪˈkwaɪər/

【動詞】必要とする、要求する | 例文:However, these detectors typically require Non-Maximum Suppression (NMS) for post-processing, which not only slows down the inference speed but also introduces hyperparameters that cause instability in both the speed and accuracy.(Introduction)

29
New cards

reshape /riˈʃeɪp/

【動詞】形状を変える、リシェイプする | 例文:O = CCFF({S3 , S4 , F5 }), where Reshape represents restoring the shape of the flattened feature to the same shape as S5 .(Method)

30
New cards

rethink /riˈθɪŋk/

【動詞】再考する、見直す | 例文:To achieve the above goal, we rethink DETRs and conduct detailed analysis of key components to reduce unnecessary computational redundancy and further improve accuracy.(Introduction)

31
New cards

select /səˈlɛkt/

【動詞】選択する | 例文:Moreover, considering that different scenarios place different emphasis on recall and accuracy, it is necessary to carefully select the appropriate NMS thresholds, which hinders the development of real-time detectors.(Introduction)

32
New cards

shows /ʃoʊz/

【動詞】示す(3人称単数現在)(基本形:show) | 例文:This shows that uncertainty-minimal query selection produces more high-quality encoder features.(Experiments)

33
New cards

suffers /ˈsʌfərz/

【動詞】(悪影響を)被る、悩まされる(基本形:suffer) | 例文:Despite its obvious advantages, DETR suffers from several problems: slow training convergence, high computational cost, and hard-to-optimize queries.(Related Work)

34
New cards

supports /səˈpɔːrts/

【動詞】対応する、サポートする(基本形:support) | 例文:In addition, RT-DETR supports flexible speed tuning by adjusting the number of decoder layers to adapt to various scenarios without retraining.(Abstract)

35
New cards

utilize /ˈjuːtəlaɪz/

【動詞】活用する、利用する | 例文:We utilize the benchmark to test the end-to-end speed of anchor-based detectors YOLOv5 [11] and YOLOv7 [38], as well as anchor-free detectors PP-YOLOE [40], YOLOv6 [16] and YOLOv8 [12] on T4 GPU with TensorRT FP16.(End-to-end Speed of Detectors)

36
New cards

visualize /ˈvɪʒuəˌlaɪz/

【動詞】可視化する | 例文:We also visualize the predictions of YOLOv8 with different NMS thresholds in Appendix.(End-to-end Speed of Detectors)

37
New cards

ablation /əˈbleɪʃən/

【名詞】アブレーション(要素実験) | 例文:We conduct an ablation study on uncertainty-minimal query selection, and the results are reported on RT-DETR-R50 with 1× configuration, Table 4.(Experiments)

38
New cards

analysis /əˈnæləsɪs/

【名詞】分析、解析 | 例文:To achieve the above goal, we rethink DETRs and conduct detailed analysis of key components to reduce unnecessary computational redundancy and further improve accuracy.(Introduction)

39
New cards

anchor /ˈæŋkər/

【名詞】アンカー(基準枠) | 例文:Particularly, DETR eliminates the hand-crafted anchor and NMS components.(Related Work)

40
New cards

approach /əˈproʊtʃ/

【名詞】手法、アプローチ/接近する | 例文:RT-DETR, along with its model scaling strategy, broadens the technical approach to real-time object detection, offering new possibilities beyond YOLO for diverse real-time scenarios.(Conclusion)

41
New cards

attention /əˈtɛnʃən/

【名詞】注意機構、アテンション | 例文:Carion et al. [4] first propose the end-to-end detector based on Transformer called DETR, which has attracted extensive attention due to its distinctive features.(Related Work)

42
New cards

backbone /ˈbækboʊn/

【名詞】バックボーン(特徴抽出部) | 例文:RT-DETR consists of a backbone, an efficient hybrid encoder, and a Transformer decoder with auxiliary prediction heads.(Method)

43
New cards

baseline /ˈbeɪslaɪn/

【名詞】基準、ベースライン(比較対象の標準手法・値) | 例文:Compared to baseline A, variant B improves accuracy by 1.9% AP and increases the latency by 54%.(Experiments)

44
New cards

benchmark /ˈbɛntʃmɑːrk/

【名詞】ベンチマーク(比較評価の基準) | 例文:We quantitatively analyze the impact of NMS on the speed and accuracy of YOLO detectors, and establish an end-to-end speed benchmark to test the end-to-end inference speed of real-time detectors.(Introduction)

45
New cards

breakthrough /ˈbreɪkθruː/

【名詞】画期的な進展、ブレークスルー | 例文:We view query initialization as a breakthrough to further improve performance.(Introduction)

46
New cards

category /ˈkætəˌɡɔri/

【名詞】カテゴリ、分類 | 例文:This process is performed iteratively until all boxes of every category have been processed.(End-to-end Speed of Detectors)

47
New cards

comparison /kəmˈpærɪsən/

【名詞】比較 | 例文:To enable a fair comparison of the end-to-end speed of various real-time detectors, we establish an end-to-end speed benchmark.(End-to-end Speed of Detectors)

48
New cards

confidence /ˈkɑːnfɪdəns/

【名詞】信頼度、確信度 | 例文:The confidence score represents the likelihood that the feature includes foreground objects.(Method)

49
New cards

configuration /kənˌfɪɡjəˈreɪʃən/

【名詞】構成、設定 | 例文:We evaluate the indicators of the variants designed in Sec. 4.2, including AP (trained with 1× configuration), the number of parameters, and the latency, Table 3.(Experiments)

50
New cards

cost /kɔːst/

【名詞】コスト、計算量・処理負担 | 例文:Nevertheless, the high computational cost limits their practicality and hinders them from fully exploiting the advantage of excluding NMS.(Abstract)

51
New cards

difficulty /ˈdɪfəkəlti/

【名詞】難しさ、困難 | 例文:Conditional DETR [27] and Anchor DETR [39] decrease the optimization difficulty of the queries.(Related Work)

52
New cards

effectiveness /ɪˈfɛktɪvnəs/

【名詞】有効性、効果 | 例文:To analyze the effectiveness of the uncertainty-minimal query selection, we visualize the classification scores and IoU scores of the selected features on COCO val2017, Figure 6.(Method)

53
New cards

embedding /ɪmˈbɛdɪŋ/

【名詞】埋め込み(特徴ベクトル) | 例文:Specifically, for the hybrid encoder, we control the width by adjusting the embedding dimension and the number of channels, and the depth by adjusting the number of Transformer layers and RepBlocks.(Experiments)

54
New cards

foreground /ˈfɔrˌɡraʊnd/

【名詞】前景(画像中の対象物体がある領域) | 例文:The confidence score represents the likelihood that the feature includes foreground objects.(Method)

55
New cards

fusion /ˈfjuːʒən/

【名詞】融合、統合 | 例文:CCFF is optimized based on the cross-scale fusion module, which inserts several fusion blocks consisting of convolutional layers into the fusion path.(Method)

56
New cards

hyperparameters /ˌhaɪpərpəˈræmətərz/

【名詞】ハイパーパラメータ(基本形:hyperparameter) | 例文:However, these detectors typically require Non-Maximum Suppression (NMS) for post-processing, which not only slows down the inference speed but also introduces hyperparameters that cause instability in both the speed and accuracy.(Introduction)

57
New cards

improvement /ɪmˈpruvmənt/

【名詞】改善、向上 | 例文:Variant C delivers a 0.7% AP improvement over B and increases the latency by 20%.(Experiments)

58
New cards

inference /ˈɪnfərəns/

【名詞】推論 | 例文:However, these detectors typically require Non-Maximum Suppression (NMS) for post-processing, which not only slows down the inference speed but also introduces hyperparameters that cause instability in both the speed and accuracy.(Introduction)

59
New cards

latency /ˈleɪtənsi/

【名詞】遅延、レイテンシ | 例文:Variant C delivers a 0.7% AP improvement over B and increases the latency by 20%.(Experiments)

60
New cards

limitation /ˌlɪməˈteɪʃən/

【名詞】制限、限界、制約 | 例文:Although the proposed RT-DETR outperforms the state-of-the-art real-time detectors and end-to-end detectors with similar size in both speed and accuracy, it shares the same limitation as the other DETRs, i.e., the performance on small objects is still inferior than the strong real-time detectors.(Limitation and Discussion)

61
New cards
localization /ˌloʊkələˈzeɪʃən/
【名詞】位置特定(局所化) | 例文:This inevitably results in encoder features with low localization confidence being selected as initial queries, thus leading to a considerable level of uncertainty and hurting the performance of DETRs.(Introduction)
62
New cards
mechanism /ˈmɛkənɪzəm/
【名詞】仕組み、機構、メカニズム | 例文:Deformable-DETR [45] accelerates training convergence with multi-scale features by enhancing the efficiency of the attention mechanism.(Related Work)
63
New cards
observation /ˌɑːbzərˈveɪʃən/
【名詞】観察、観測(その結果) | 例文:To verify this observation, we leverage YOLOv5 [11] (anchor-based) and YOLOv8 [12] (anchor-free) for analysis.(End-to-end Speed of Detectors)
64
New cards
optimization /ˌɑptəməˈzeɪʃən/
【名詞】最適化 | 例文:Conditional DETR [27] and Anchor DETR [39] decrease the optimization difficulty of the queries.(Related Work)
65
New cards
performance /pərˈfɔːrməns/
【名詞】性能、パフォーマンス | 例文:Compared to previously advanced real-time object detectors, our RT-DETR achieves state-of-the-art performance.(Introduction)
66
New cards
post-processing /poʊst ˈprɑsɛsɪŋ/
【名詞】後処理 | 例文:However, these detectors typically require Non-Maximum Suppression (NMS) for post-processing, which not only slows down the inference speed but also introduces hyperparameters that cause instability in both the speed and accuracy.(Introduction)
67
New cards
quality /ˈkwɑləti/
【名詞】品質、質 | 例文:Then, we propose the uncertainty-minimal query selection to provide high-quality initial queries to the decoder, thereby improving accuracy.(Abstract)
68
New cards
recall /rɪˈkɔːl/
【名詞】再現率(正解のうち検出できた割合を示す指標) | 例文:Moreover, considering that different scenarios place different emphasis on recall and accuracy, it is necessary to carefully select the appropriate NMS thresholds, which hinders the development of real-time detectors.(Introduction)
69
New cards
scheme /skiːm/
【名詞】方式、仕組み、計画 | 例文:DAB-DETR [23] and DN-DETR [17] further improve performance by introducing the iterative refinement scheme and denoising training.(Related Work)
70
New cards
threshold /ˈθrɛʃoʊld/
【名詞】しきい値、閾値 | 例文:Specifically, the boxes with scores below the confidence threshold are directly filtered out, and whenever the IoU of any two boxes exceeds the IoU threshold, the box with the lower score will be discarded.(End-to-end Speed of Detectors)
71
New cards
tuning /ˈtunɪŋ/
【名詞】調整、チューニング | 例文:In addition, RT-DETR supports flexible speed tuning by adjusting the number of decoder layers to adapt to various scenarios without retraining.(Abstract)
72
New cards
variant /ˈvɛriənt/
【名詞】変種、派生版 | 例文:Compared to baseline A, variant B improves accuracy by 1.9% AP and increases the latency by 54%.(Experiments)
73
New cards
accurate /ˈækjərət/
【形容詞】正確な、精度の高い | 例文:Quantitative results further demonstrate that the uncertainty-minimal query selection provides more features with accurate classification and precise location for queries, thereby improving the accuracy of the detector (cf. Sec. 5.3).(Experiments)
74
New cards
advanced /ədˈvænst/
【形容詞】高度な、先進的な | 例文:We build RT-DETR in two steps, drawing on the advanced DETR: first we focus on maintaining accuracy while improving speed, followed by maintaining speed while improving accuracy.(Abstract)
75
New cards
anchor-free /ˈæŋkər ˈfri/
【形容詞】アンカーフリーの(アンカーを用いない) | 例文:According to the results (cf. Table 2), we conclude that anchor-free detectors outperform anchor-based detectors with equivalent accuracy for YOLO detectors because the former require less NMS time than the latter.(End-to-end Speed of Detectors)
76
New cards
auxiliary /ɔɡˈzɪljəri/
【形容詞】補助的な(補助ヘッドなど) | 例文:RT-DETR consists of a backbone, an efficient hybrid encoder, and a Transformer decoder with auxiliary prediction heads.(Method)
77
New cards
beneficial /ˌbɛnəˈfɪʃəl/
【形容詞】有益な、効果のある | 例文:For the former, we observe that although the introduction of multi-scale features is beneficial in accelerating the training convergence [45], it leads to a significant increase in the length of the sequence feed into the encoder.(Introduction)
78
New cards
consistent /kənˈsɪstənt/
【形容詞】一貫した、矛盾のない | 例文:We test the speed on T4 GPU with TensorRT FP16, and the input and preprocessing remain consistent.(End-to-end Speed of Detectors)
79
New cards
continuous /kənˈtɪnjuəs/
【形容詞】連続した、連続的な | 例文:Through years of continuous development, the YOLO detectors have outperformed other one-stage object detectors [21, 24] and become the synonymous with the real-time object detector.(Related Work)
80
New cards
corresponding /ˌkɔːrəˈspɑːndɪŋ/
【形容詞】対応する | 例文:The hyperparameters and the corresponding results are shown in Table 1.(End-to-end Speed of Detectors)
81
New cards
cross-scale /ˈkrɔːs skeɪl/
【形容詞】スケールをまたぐ、クロススケール | 例文:Specifically, we design an efficient hybrid encoder to expeditiously process multi-scale features by decoupling intra-scale interaction and cross-scale fusion to improve speed.(Abstract)
82
New cards
detailed /dɪˈteɪld/
【形容詞】詳細な、詳しい | 例文:To achieve the above goal, we rethink DETRs and conduct detailed analysis of key components to reduce unnecessary computational redundancy and further improve accuracy.(Introduction)
83
New cards
evident /ˈɛvɪdənt/
【形容詞】明らかな、明白な | 例文:The same conclusion can be drawn from the density curves, where the gap between purple and green is most evident in the top right of the figure.(Experiments)
84
New cards
experimental /ɪkˌspɛrəˈmɛntəl/
【形容詞】実験的な、実験の | 例文:More experimental results are provided in the Appendix.(Introduction)
85
New cards
extensive /ɪkˈstɛnsɪv/
【形容詞】広範な、大規模な | 例文:Carion et al. [4] first propose the end-to-end detector based on Transformer called DETR, which has attracted extensive attention due to its distinctive features.(Related Work)
86
New cards
flexible /ˈflɛksəbəl/
【形容詞】柔軟な | 例文:In addition, RT-DETR supports flexible speed tuning by adjusting the number of decoder layers to adapt to various scenarios without retraining.(Abstract)
87
New cards
hand-crafted /ˈhænd ˈkræftəd/
【形容詞】手作業で設計された(特徴量など) | 例文:Particularly, DETR eliminates the hand-crafted anchor and NMS components.(Related Work)
88
New cards
high-quality /ˈhaɪ ˈkwɑləti/
【形容詞】高品質の | 例文:Then, we propose the uncertainty-minimal query selection to provide high-quality initial queries to the decoder, thereby improving accuracy.(Abstract)
89
New cards
inappropriate /ˌɪnəˈproʊpriət/
【形容詞】不適切な、当てはまらない | 例文:The results show that inappropriate confidence thresholds lead to significant false positives or false negatives by the detector.(End-to-end Speed of Detectors)
90
New cards
intra-scale /ˈɪntrə skeɪl/
【形容詞】スケール内(同一解像度内) | 例文:Specifically, we design an efficient hybrid encoder to expeditiously process multi-scale features by decoupling intra-scale interaction and cross-scale fusion to improve speed.(Abstract)
91
New cards
iterative /ˈɪtəreɪtɪv/
【形容詞】反復的な、繰り返しの | 例文:DAB-DETR [23] and DN-DETR [17] further improve performance by introducing the iterative refinement scheme and denoising training.(Related Work)
92
New cards
lightweight /ˈlaɪtˌweɪt/
【形容詞】軽量な(モデルのパラメータ数・計算量が小さい) | 例文:The proposed RT-DETR at different scales preserves decoders homogeneous to other DETRs, which makes it possible to distill our lightweight detector with high accuracy pre-trained large DETR models.(Limitation and Discussion)
93
New cards
minimal /ˈmɪnəməl/
【形容詞】最小限の、最小の | 例文:We observe that removing a few decoder layers at the end has minimal effect on accuracy, but greatly enhances inference speed (cf. Sec. 5.4).(Method)
94
New cards
multi-scale /ˈmʌlti skeɪl/
【形容詞】マルチスケール(複数解像度) | 例文:The efficient hybrid encoder transforms multi-scale features into a sequence of image features through intra-scale feature interaction and cross-scale feature fusion (cf. Sec. 4.2).(Method)
95
New cards
necessary /ˈnɛsəˌsɛri/
【形容詞】必要な | 例文:Moreover, considering that different scenarios place different emphasis on recall and accuracy, it is necessary to carefully select the appropriate NMS thresholds, which hinders the development of real-time detectors.(Introduction)
96
New cards
numerous /ˈnuːmərəs/
【形容詞】多数の、数多くの | 例文:These advanced real-time detectors produce numerous overlapping boxes and require NMS post-processing, which slows down their speed.(Related Work)
97
New cards
reasonable /ˈrizənəbəl/
【形容詞】妥当な、合理的な | 例文:The YOLO series has become the most popular framework for real-time object detection due to its reasonable trade-off between speed and accuracy.(Abstract)
98
New cards
scalable /ˈskeɪləbəl/
【形容詞】拡張可能な、スケーラブルな | 例文:Lighter RT-DETRs can be designed by applying other smaller (e.g., ResNet18/34) or scalable (e.g., CSPResNet [40]) backbones with scaled encoder and decoder.(Method)
99
New cards
semantic /səˈmæntɪk/
【形容詞】意味的な(セマンティック) | 例文:Intuitively, high-level features that contain rich semantic information about objects are extracted from low-level features, making it redundant to perform feature interaction on the concatenated multi-scale features.(Method)
100
New cards
sensitive /ˈsɛnsɪtɪv/
【形容詞】敏感な、影響を受けやすい | 例文:We sample values from 0.001 to 0.25 as confidence thresholds to count the number of remaining boxes of the two detectors and plot them on a bar graph, which intuitively reflects that NMS is sensitive to its hyperparameters, Figure 2.(End-to-end Speed of Detectors)