RT-DETR important words 30 (3-part)

facilitates /fəˈsɪlɪˌteɪts/

【動詞】促進する、容易にする（基本形：facilitate） ｜ 例文：The reason is that applying the self-attention operation to high-level features with richer semantic concepts captures the connection between conceptual entities, which facilitates the localization and recognition of objects by subsequent modules.（Method）

formulate /ˈfɔːrmjəleɪt/

【動詞】定式化する ｜ 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）

denote /dɪˈnoʊt/

【動詞】（記号などが）〜を表す、示す ｜ 例文：where Ŷ and Y denote the prediction and ground truth, Ŷ = {ĉ, b̂}, ĉ and b̂ represent the category and bounding box respectively, X̂ represent the encoder feature.（Method）

integrate /ˈɪntɪɡreɪt/

【動詞】統合する、組み込む ｜ 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）

discarded /dɪˈskɑrdɪd/

【動詞】捨てる、除外する（基本形：discard） ｜ 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）

accelerating /ækˈsɛləˌreɪtɪŋ/

【動詞】加速する、促進する（基本形：accelerate） ｜ 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）

demonstrate /ˈdɛmənˌstreɪt/

【動詞】実証する、示す ｜ 例文：However, their high computational cost prevents them from meeting real-time detection requirements, so the NMS-free architecture does not demonstrate an inference speed advantage.（Introduction）

verify /ˈvɛrəˌfaɪ/

【動詞】検証する、確認する ｜ 例文：To verify this opinion, we perform the intra-scale interaction only on S5 in variant D, and the experimental results are reported in Table 3 (see row DS5 ).（Method）

uncertainty /ʌnˈsɜːrtənti/

【名詞】不確実性 ｜ 例文：Subsequently, the uncertainty-minimal query selection is employed to select a fixed number of encoder features to serve as initial object queries for the decoder (cf. Sec. 4.3).（Method）

discrepancy /dɪˈskrɛpənsi/

【名詞】不一致、乖離（分類と位置の予測分布の差） ｜ 例文：Specifically, the feature uncertainty U is defined as the discrepancy between the predicted distributions of localization P and classification C in Eq. (2).（Method）

redundancy /rɪˈdʌndənsi/

【名詞】冗長性（無駄な重複） ｜ 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）

convergence /kənˈvɝdʒəns/

【名詞】収束（学習の収束） ｜ 例文：Despite its obvious advantages, DETR suffers from several problems: slow training convergence, high computational cost, and hard-to-optimize queries.（Related Work）

bottleneck /ˈbɑtəlˌnɛk/

【名詞】ボトルネック（全体の速度を制限する部分） ｜ 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）

likelihood /ˈlaɪkliˌhʊd/

【名詞】尤度、〜である可能性・確からしさ ｜ 例文：The confidence score represents the likelihood that the feature includes foreground objects.（Method）

instability /ˌɪnstəˈbɪləti/

【名詞】不安定さ ｜ 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）

indicators /ˈɪndəˌkeɪtərz/

【名詞】指標（基本形：indicator） ｜ 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）

interaction /ˌɪntərˈækʃən/

【名詞】相互作用（特徴間の相互作用） ｜ 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）

considerable /kənˈsɪdərəbəl/

【形容詞】かなりの、相当な ｜ 例文：Based on the analysis, the current query selection lead to a considerable level of uncertainty in the selected features, resulting in sub-optimal initialization for the decoder and hindering the performance of the detector.（Method）

subsequent /ˈsʌbsəkwənt/

【形容詞】後続の、その後の ｜ 例文：The reason is that applying the self-attention operation to high-level features with richer semantic concepts captures the connection between conceptual entities, which facilitates the localization and recognition of objects by subsequent modules.（Method）

adjacent /əˈdʒeɪsənt/

【形容詞】隣接する ｜ 例文：Furthermore, we observe that the difference in accuracy between adjacent decoder layers gradually decreases as the index of the decoder layer increases.（Experiments）

distinctive /dɪˈstɪŋktɪv/

【形容詞】特徴的な、際立った ｜ 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）

vanilla /vəˈnɪlə/

【形容詞】バニラ（改変されていない素の、基本形の） ｜ 例文：To expeditiously process multi-scale features, we design an efficient hybrid encoder to replace the vanilla Transformer encoder, which significantly improves inference speed by decoupling the intra-scale interaction and cross-scale fusion of features with different scales.（Related Work）

learnable /ˈlɜːrnəbəl/

【形容詞】学習可能な ｜ 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）

sub-optimal /ˌsʌbˈɑːptəməl/

【形容詞】最適ではない、準最適な ｜ 例文：Based on the analysis, the current query selection lead to a considerable level of uncertainty in the selected features, resulting in sub-optimal initialization for the decoder and hindering the performance of the detector.（Method）

precise /prɪˈsaɪs/

【形容詞】正確な、精密な ｜ 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）

quantitatively /ˈkwɑːntəˌteɪtɪvli/

【副詞】定量的に ｜ 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）

simultaneously /ˌsaɪməlˈteɪniəsli/

【副詞】同時に ｜ 例文：Nevertheless, the detector are required to simultaneously model the category and location of objects, both of which determine the quality of the features.（Method）

inevitably /ɪnˈɛvətəbli/

【副詞】必然的に、避けられず ｜ 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）

significantly /sɪɡˈnɪfɪkəntli/

【副詞】大幅に、著しく ｜ 例文：To expeditiously process multi-scale features, we design an efficient hybrid encoder to replace the vanilla Transformer encoder, which significantly improves inference speed by decoupling the intra-scale interaction and cross-scale fusion of features with different scales.（Related Work）

explicitly /ɪkˈsplɪsətli/

【副詞】明示的に ｜ 例文：To address this problem, we propose the uncertainty minimal query selection scheme, which explicitly constructs and optimizes the epistemic uncertainty to model the joint latent variable of encoder features, thereby providing high-quality queries for the decoder.（Method）

light /laɪt/

【形容詞】軽い、わずかな；（in light of 〜 で）〜を考慮して ｜ 例文：In light of the computational bottleneck in the encoder, we propose an efficient hybrid encoder to accelerate inference.（Method）