nopperl/arxiv-image-text · Datasets at Hugging Face

dedup-isc-ft-v107-score float64 0.32 1	uid stringlengths 32 32	text stringlengths 0 32.5k	paper_id stringlengths 1 14	original_image_filename stringlengths 5 222
0.502961	9b5cb981e9754fc5bd63fad1dd0edd8a	(a) Superimposed images of a mucosalivary spray emerging from the mouth of a 3D printed mannequin face obtained at 1000 fps over 4 seconds. Color bar indicates progression of time from white t=0 s (white) to t=4 s (orange). The mucosaliva is marked with a fluorescent tracer and deposits over time onto a horizontal surface. (b) The light intensity, I(x,y), corresponding to mucosaliva deposited on the surface in front of the head after n_s = 5 (top) and n_s =20 (bottom) expirations, or sprays (H = 0.5 m). Images are taken approximately 10 seconds after the last spray. A deposition lobe directed along the symmetry axis in front of the face can be observed. The laser line is set at y=0 m. (c) Images of deposited mucosalivary fluid after n_s = 20 sprays illuminated by a 1 cm wide laser light sheet at distance from source of (i) x = 0.35 m, (ii) x=0.5 m, and (iii) x=1.0 m.	2012.14268	Fig1_schematic.jpg
0.450382	d144b50667ba4f099c9fd09c8fb4438d	(a) Side view snapshot of a mucosalivary spray at t=0.35 s with a cloud threshold boundary (green line), where ∘ marks the cloud geometric center, and ▹ and ▿ mark the furthest x extent and the lowest z extent of the boundary, respectively. The green square and diamond indicates the furthest and lowest visible droplets in the x- and z-directions, respectively. (b) Tracked boundary of the cloud over time and the cone angle θ_s. The geometric center of the cloud asymptotes towards x=0.4 m, whereas individual droplets separate from the cloud travel further. The initial propagation speed is u_0=5 m/s. (d) The settling speed for the center, boundary, and individual droplet is found to be u_sc = 0.5 m/s, u_sb = 0.8 m/s, and u_sd = 2.0 m/s, respectively. Each point represents the mean over n_t = 10 trials with error bars of one standard deviation.	2012.14268	Fig2_cloudBoundary.jpg
0.506667	f1a737ec43944655b2bff1940354e298	Distribution profile of mucosalivary droplets deposited on a horizontal surface along x-axis from heights of (a) H=0.5 m, (b) H=1 m, and (c) H=1.5 m (n_s = 20). Insets are the respective semi-log plots. The noise floor is ϕ_noise=4.4×10^-6, which is three orders of magnitude less that the maximum signal. A total of n_t=12 trials are conducted for each height. Data points are binned with a bin size of 0.1 m. Distribution width σ_x ≈ 0.5 m for all heights. Black line is a fitted exponential function. (d) Maximum amount of material, ϕ_m decreases with height. (e) The mean of the distribution, x̅, decreases linearly with increasing height. (f) Distance x^(98%) at which 98% of the material is deposited decreases with H. Each point represents the mean over n_t=12 trials with error bars of one standard deviation.	2012.14268	Fig3_I-r_varyH.jpg
0.444621	75bfe178c59a4739abf480b1caaf4cbb	(a) The time evolution of the volume of droplets over a range of x=[0.7, 0.8] m. Droplets are expelled from a height of H=1.5 m. Volumes are self-normalized by the maximum volume such that ϕ/ϕ_∞. Time is normalized by the gravitational settling time, t_g=√(2H/g)∼ 0.55 s. Solid blue line is the average of n_t=5 trials (light gray lines). (b) The landing time, τ, of droplets at H=1.5 m. The time it takes to reach a percentage of ϕ_∞, τ, is plotted for x=40-70 cm. At a given x, the time to reach 25%, 50%, 75%, 90%, and 95% of ϕ_∞ goes from left to right. Error bars are one standard deviation from the mean.	2012.14268	Fig4_landingTime.jpg
0.478339	6d7b466cf71e4739a4372f309efb7b60	Trajectories T1-T4 of individual droplets corresponding to some of the brighter traces in Fig. <ref>(a). (a) Some droplets travel with pure projectile motion as shown by good agreement with Eq. (<ref>). (b) Trajectories T3 and T4 deviate from projectile motion due to air drag. The initial conditions are for T1: θ=-0.7^∘, u_0=2.25 m/s; T2: θ=25^∘, u_0=1.95 m/s; T3: θ=45^∘, u_0=1.8 m/s; and T4: θ=-20^∘, u_0=2.0 m/s.	2012.14268	Fig5_dropTrajectories.jpg
0.430411	93f996ab03f041f2b757382eee4d1ec2	(a) The average 2D projected area of the cloud, A_c, over time increases linearly after t=0.2 s (blue circles) with a slope of Ȧ_c=0.2 m^2/s. Error bars are one standard deviation. (b) The average settling speed of the cloud relative to the average effective diameter, D_c=√(4A_c/π) (blue circles) after t>0.2 s. The lines are implicit calculations of the terminal velocity as a function of discrete diameters. Our data falls within the the calculations in which cloud density, ρ_c, is between 0.5-2.5% greater than the density of air, ρ_a. An estimated density ratio of ρ_c/ρ_a= 1.011 agrees well with the experimental data. Gray circles indicate the the cloud growth for all n_t=10 trials. One standard deviation for diameter and velocity is shown in lower right corner.	2012.14268	Fig6_settlingSpeed.jpg
0.423309	10d769db810a4153ba6e578475ed5fc7	(a) Comparing the distribution profile of mucosalivary droplets for no mask vs with mask from a height of 1.5 m. (b) Same comparison as above on semilog scale. n_t=5 trials for both cases. (c) Switching to a N95 mask reduces the dispersal to within the noise floor of our measurements. Wearing a standard non-medical or N95 mask significantly reduces the total volume by at least 99%, and reduces the peaks recorded by at least 99.5%. Noise floor is ϕ_noise=4.4×10^-6, as shown in gray.	2012.14268	Fig7_mask-noMask.jpg
0.510738	bfce0fbd79b8414ea2e71865035fbcb9	Proposed manga translation framework. N' represents the translation of a source sentence N. Mitsuki Kuchitaka	2012.14271	context_translation.jpg
0.433214	6f29ff897b264f80ae8b244b49e67872	Results of fully automatic manga translation from Japanese to English and Chinese. Masami Taira, Syuji Takeya	2012.14271	e2e_example_multilang.jpg
0.410126	5ed16702f5644ed4a5e0f128349d3501	Given a manga page, our system automatically translates the texts on the page into English and replaces the original texts with the translated ones. Mitsuki Kuchitaka.	2012.14271	eg_first.jpg
0.435301	b270af186c7645f18b4d48a1bb607687	Outputs of the sentence-based (center) and frame-based (right) models. The values after H and B are respectively the human evaluation and BLEU scores for each page. Mitsuki Kuchitaka	2012.14271	eg_swapped.jpg
0.446362	bfc80347af3b402ba541859cbe570449	Translations output by the sentence-based model without (i) and with visual information (ii). By overwriting the character face in the input image (a) with a male face (b), the pronouns in the translation results (iii) are also changed. Nako Nameko	2012.14271	eg_visual.jpg
0.416672	281e7fb1a585463b9c45aca52138f14d	Translation results with and without visual features. Miki Ueda, Satoshi Arai.	2012.14271	eg_visual_2.jpg
0.447479	a77928a4d5f7455a976aaa51d64ac71d	Proposed framework of parallel corpus construction.	2012.14271	framework_collecting_groundtruth.jpg
0.448261	fc4a184944424b65b7bb4b79f1dc1cbf	GUI of the user study. Mitsuki Kuchitaka	2012.14271	gui_v2.jpg
0.464935	2812e6dc2d524edc95d507899cc94ce4	Procedures of rendering the training data.	2012.14271	synth_generation.jpg
0.411989	5dbed917ef944b4e8e575e61a8ed4fec	Term definition for manga text. Mitsuki Kuchitaka	2012.14271	term_definition.jpg
0.407008	8234aef7ad354add811d5f601fdc4fc2	Rules-based text-line detection.	2012.14271	text_detection_recognition_v4.jpg
0.419667	e80ede6d45f74664846e8f510ffacb4a	Generation of textline annotations.Yasuyuki Ohno	2012.14271	textline_detector.jpg
0.423891	5fb41be6219e42abaff7889f4841393b	Sample images with main upper-body line and clothing brightness value (bv). In the full dataset, these images are (from left to right) at brightnessvalue percentiles 0, 33, 66, and 100.	2012.14274	color_quantiles.jpg
0.441075	f4bcc5c04a724335869be897993f0cc0		2012.14274	photoshop_1.jpg
0.465732	d41d18d1e3fa489ca752e9d6f22b0735		2012.14274	photoshop_1_dark.jpg
0.448865	d91c63b78d014c92a1237be942d3340f		2012.14274	photoshop_1_light.jpg
0.430381	65490ef5844c4127a253fd43e42e6792		2012.14274	photoshop_2.jpg
0.494076	782ea4dfa61c469e9aac02e7229bbbe8		2012.14274	photoshop_2_dark.jpg
0.500136	6e8d04d92249427eb1499f9118886634		2012.14274	photoshop_2_light.jpg
0.468993	be34dd5549834cf5b6ef40b792fc9dca	BMI	2012.14274	plot_dist_bmi_pairwise_est_clean_dark_light.jpg
0.430929	0c934508b23940269eb0f2a6490124e2	BMI	2012.14274	plot_dist_bmi_pairwise_est_clean_dark_striped.jpg
0.463908	7e753dda0f0e471ba96756d74ad1aeb7		2012.14274	plot_dist_bmi_pairwise_est_clean_dl.jpg
0.434811	c56a27933cff4dc8a0dbd7c2d2f3dc6b	BMI	2012.14274	plot_dist_bmi_pairwise_est_clean_light_striped.jpg
0.403667	f3fcb96759fa48a095c6d43d3026a1d1	BMI	2012.14274	plot_dist_bmi_pairwise_est_dirty_dark_light.jpg
0.459425	be81486941ea4551a95c33323b82a548	BMI	2012.14274	plot_dist_bmi_pairwise_est_dirty_dark_striped.jpg
0.420217	69f2a23b42c8442e8a08278544a624ed		2012.14274	plot_dist_bmi_pairwise_est_dirty_dl.jpg
0.447394	e342276c8c644ea8b90c226079d4fd92	BMI	2012.14274	plot_dist_bmi_pairwise_est_dirty_light_striped.jpg
0.463984	712406bf61ba4654b05368887369fd01		2012.14274	plot_dist_bmi_pairwise_est_dirty_ls.jpg
0.434953	138ae42d07584ac28e1d7fbb5bb345ef	Height	2012.14274	plot_dist_h_pairwise_est_clean_dark_light.jpg
0.43767	910b67221eae44579fd24353bfe2ddbf	Height	2012.14274	plot_dist_h_pairwise_est_clean_dark_striped.jpg
0.4223	65ae1c03cebc4066bc613ae707aec937		2012.14274	plot_dist_h_pairwise_est_clean_dl.jpg
0.426563	68a15bc29a58419aa549b7be24546b1a	Height	2012.14274	plot_dist_h_pairwise_est_clean_light_striped.jpg
0.453957	6218852ea5fa4943a37d78527ad69499	Height	2012.14274	plot_dist_h_pairwise_est_dirty_dark_light.jpg
0.462967	f1e33ddbf2804b4a96db455901a75ebd	Height	2012.14274	plot_dist_h_pairwise_est_dirty_dark_striped.jpg
0.43535	ae8ec2b2143c4ea9b220b0872445b3e5		2012.14274	plot_dist_h_pairwise_est_dirty_dl.jpg
0.499694	03331079a5cb4fd6a34928514b9b3afd	Height	2012.14274	plot_dist_h_pairwise_est_dirty_light_striped.jpg
0.416175	7302fb6e37054ec4a2315d06dfe661c0		2012.14274	plot_dist_h_pairwise_est_dirty_ls.jpg
0.427711	5d511571e8954e31ab252067dbce4eb6	Weight	2012.14274	plot_dist_w_pairwise_est_clean_dark_light.jpg
0.52511	0beed24f527a4da890dc068599d7002e	Weight	2012.14274	plot_dist_w_pairwise_est_clean_dark_striped.jpg
0.476577	4f4fdfbb263c4a72b899fb7782f31775		2012.14274	plot_dist_w_pairwise_est_clean_dl.jpg
0.456272	098acac093f1402a8fb8c30717397acf	Weight	2012.14274	plot_dist_w_pairwise_est_clean_light_striped.jpg
0.467213	7f71c1e710dc41ce96ef099d78cf1bfb	Weight	2012.14274	plot_dist_w_pairwise_est_dirty_dark_light.jpg
0.454529	d1c8c6fffa6e4931b22d117439cb0741	Weight	2012.14274	plot_dist_w_pairwise_est_dirty_dark_striped.jpg
0.451635	f6f278f099ea4594983015709cd4b72d		2012.14274	plot_dist_w_pairwise_est_dirty_dl.jpg
0.442856	7709625cc7cf4e2bb5719a4249d50e7f	Weight	2012.14274	plot_dist_w_pairwise_est_dirty_light_striped.jpg
0.476604	2e1df82ee6b046649a957882ad80312b		2012.14274	plot_dist_w_pairwise_est_dirty_ls.jpg
0.452696	fccd29bd76844eb4948f482ed89f7abc		2012.14274	plot_scatter_bmi_pairwise_comp_clean_dl.jpg
0.463921	71925f45158f4ef49334176de9b78403		2012.14274	plot_scatter_bmi_pairwise_comp_dirty_dl.jpg
0.398461	bc0fe06e9209484593fdcd79c2d85122		2012.14274	plot_scatter_bmi_pairwise_comp_dirty_ls.jpg
0.454578	459e15247f3d4d7abf05e0a9a274f3c2		2012.14274	plot_scatter_h_pairwise_comp_clean_dl.jpg
0.436289	e7bfe34fa9064bd69cb63598a2caf3de		2012.14274	plot_scatter_h_pairwise_comp_dirty_dl.jpg
0.439047	bf16322260c54e54a103447f64a5ed9d		2012.14274	plot_scatter_h_pairwise_comp_dirty_ls.jpg
0.462887	e31c575f11994a10b34eb7d58031c971		2012.14274	plot_scatter_w_pairwise_comp_clean_dl.jpg
0.416554	cbd340d9274c40e9affffe112038ab79		2012.14274	plot_scatter_w_pairwise_comp_dirty_dl.jpg
0.440359	3b862f22de2941fab86f9a15f13a20f5		2012.14274	plot_scatter_w_pairwise_comp_dirty_ls.jpg
0.480596	b2dd42aadcd14cdc8bf035e623ef6ec1	Brightness vector	2012.14274	plot_stripes_colorline.jpg
0.479454	f319087f5eb84e279086253e5f356ac0	Fourier spectrum	2012.14274	plot_stripes_colorline_fourier.jpg
0.426065	e1e9401b8b5e403a9e786b4f5d753510	Smoothed brightness vector	2012.14274	plot_stripes_colorline_smooth.jpg
0.460572	2c0081b28460413590b067dda6ddce9d	Image	2012.14274	plot_stripes_sample.jpg
0.378165	5408f317bf6748df87d0cfbe8db11c56	Schematic of direction discovery and image transformation. The method is broadly useful with different generators and could be augmented with an inverter to take real-world images as inputs.	2012.14283	z_L1_schematic_updated.jpg
0.437045	74ffe99ae51840af8a9295044901cb91	Learning by ignoring. Along the solid arrows, predictions are made and train/validation losses are calculated. Along the dotted arrows, gradients are calculated and parameters (including ignoring variables and network weights) are updated.	2012.14288	archi-2-lbi.jpg
0.432753	f199dca50dfb482f80960eac6a8b1b31	Illustration of learning by ignoring. Void circles denote ignored pretraining data examples. Given a set of intermediately selected pretraining examples, they are used to pretrain a model, which is then finetuned and validated. The validation performance provides guidance on what pretraining examples should be ignored in the next round of learning. This process iterates until convergence.	2012.14288	illustration-lbi.jpg
0.518656	69ad135e33fe486d9e03c66b66ad6695	This irresistable Far Side cartoon by the magnificent Gary Larson does not quite express the premise of this BoF. To illustrate that the Users, Developers and Managers of scientific software are holding each other back by the way they are organized, the formation of the groundbound waddlers should of course be different from the lofty flyers. But we trust that you get the point anyway.	2012.14290	B11-143_f1.jpg
0.453781	e6ac24932fe24bc19e53666063d28114	The evolutionary behaviour of the deceleration parameter q versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig10.jpg
0.53439	3ac59a068ce546aa9da402ed5a44be12	The evolutionary behaviour of the Barrow energy density parameter Ω_D versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig11.jpg
0.438543	b54096c049ea4fb5a28a60cca43bbd43	The evolutionary behaviour the squared sound speed v_s^2 versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig12.jpg
0.466106	086f8ed482c44a05a093de672a2ed21f	The evolutionary behaviour of equation of state parameter ω_D versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig21.jpg
0.491952	f095dec1252b4110b0a50b353e976deb	The evolutionary behaviour of the deceleration parameter q versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig22.jpg
0.462113	73159a5b7e7641e1a643b3e80f365256	The evolutionary behaviour of the Barrow energy density parameter Ω_D versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig23.jpg
0.431694	877873fe8f214d17936109a0892934cb	The evolutionary behaviour of the squared sound speed v^2_s versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig24.jpg
0.506608	c36264b7ecfb47cba5b70203d9c0d419	The evolutionary behaviour of equation of state parameter ω_D versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.	2012.14291	fig9.jpg
0.505152	dc231e7a3fda467797c7e289032fc115	Stability potential (<ref>) for a symmetric kink with power-law asymptotics: the potential is V(φ) = 1/2cos^4φ, the corresponding kink is φ_ K^(x) = arctg x.	2012.14298	fig1.jpg
0.457744	c8eb67f114444911ad784943e44bbab5	The spectrum according to eq. p5_1 for a string with N=40 sections, x=0.3 and L=1. The dots denote the eigenvalues. For N→∞ or L→∞ these approach the solid line.	2012.14301	fig1_v2.jpg
0.524014	7f031df3a7754a61a3a9da2219e762bf	The vacuum energy E_0, E01, E2, for a composite string as a function of the ratio x, 2_3, of the string tensions for several numbers N of sections of the string. N increases from top to bottom (color online). The length of the string is L=1.	2012.14301	fig2_v2.jpg
0.428764	3d13a6f45cb149a7a650b7a1bc1f76bc	The limiting slope w(x) of the vacuum energy4.6 for N→∞ as a function of the ratio x.	2012.14301	fig3_v2.jpg
0.635904	14cbb49478ca434ebd23f8bbf806c799	Drawing of one detector half in the rϕ view, showing the arrangement of ladders in the four detector layers.	2012.14304	BPIX-Sensor-Module-Positions2.jpg
0.430248	f026a15fd8ac41128aff54e8ee276690	One half of the detector.	2012.14304	BPIXFinal.jpg
0.398587	76be6be848234491bcdd7c9d4e016b69	Picture of a service half-cylinder during construction. The service half-cylinder is placed on a rotatable mandrel. Segments A and B are made from Airex foam, while Segment C is made from carbon fiber plates.	2012.14304	BPIXSTUZH.jpg
0.404951	ae01aae6a0ee42fcb711960e0d3928cd	Thermal image of an disk during quality control. An infrared camera observed the disk through a porthole in the test enclosure and was programmed such that the allowed temperature range, between -20.0 and -10.5 , was rendered in the color spectrum from purple to white, while temperatures outside of the allowed range were rendered green. The blade surfaces of the disk, except for the areas where the heaters and co-cured reflective pads are located (corresponding to the green large and small rectangles), appear to be within the acceptable range.	2012.14304	FPIX_disk_thermal.jpg
0.395259	e691d7bc995e4e449f89e9840d7d836c	Cluster hit efficiency as a function of instantaneous luminosity measured using tracks from collisions in 2018.	2012.14304	Hitefficiency_new.jpg
0.444936	ebc7f37af36a4a13be02753dd861c53f	Single-pixel hit efficiency versus X-ray hit rate measured for double columns in a ROC of a L2 module. The dashed lines indicate the values at which the single-pixel hit efficiency per double column is determined during the module qualification and cover the range of expected hit rates in L2-L4. The solid red line is a fit to the data points using a polynomial function.	2012.14304	L2efficiency.jpg
0.416283	26c24bc5270c490eaff153db2bceee48	Picture of the L2 mechanics during production.	2012.14304	L2mechanics.jpg
0.454154	854006f43ca9410baccf420979979e01	Layout of the compared to the original detector layout, in longitudinal view.	2012.14304	Layout.jpg
0.434217	627d2b1369f44ffd928f6bfb2527b132	Drawings of the pixel detector modules for L1 (left), L2-4 (middle), and the detector (right).	2012.14304	Modules.jpg
0.471011	22b04715db8d4f40aa6bd4d7deb40ec5	Pulse height measured in one pixel as a function of amplitude of the internal calibration signal in units of . The slope of the curve below saturation is fitted with a linear function.	2012.14304	PHvsVcal.jpg
0.413213	edd7deae3d124664a0f1c3d823142038	DC photocurrent measured at the FED receiver as a function of the setting of the laser bias current on the POH. The measured values for one FED (24 optical links) are shown. Each curve corresponds to one optical link. The bias current is programmable through the interface in steps of 0.45 mA per setting. The setting is adjusted such that the output current is above 0.1 mA (indicated by the horizontal line). The spread between optical links is caused by differences in the quality of the optical connection.	2012.14304	POHBias.jpg
0.456981	2914af05ec144a23a46a697b18372b0d	Drawing of the Phase-1 and detectors together with the service half-cylinders that hold the readout and control circuits as well as power and cooling lines.	2012.14304	PixelSystem.jpg
0.462632	226942498b3545959c7d334ebf13acc5	Picture of one side of the CMS detector with the endcaps opened and the temporary platform for the pixel detector installation set up. Pictures used by permission from CERN.	2012.14304	SSP_99.jpg
0.580188	fe7b2d04d58e47ce931565c5f9a070a5	Calibration of one ROC on a pixel detector module with monochromatic X-rays. The measured pulse height is plotted against the expected number of electrons from the K_α-transitions in the different target materials, as given in Ref. <cit.>. The pulse height in units of the internal calibration signal has a linear dependence on the number of electrons expected to be collected from different target materials. The red line shows a fit with a linear function to the data points.	2012.14304	XrayCalibrationChipResults.jpg
0.409497	697b5fd344b246dd9175b7c5586937f3	Drawing of a service half-cylinder.	2012.14304	bpixSTDrawingNew.jpg
0.455869	1f0d55c4b82944d8925192dc0a2c667a	Measured single-pixel hit efficiency for the as a function of pixel hit rate in the 200 MeV proton beam at PSI. The version of the used in the CMS experiment in operation in 2017 and 2018 is compared to the revised version of the chip that will be used in the replacement of L1 in Run 3. The yellow and green points labeled "with resets" correspond to the data-taking mode in which every trigger signal was preceded by a reset signal.	2012.14304	effproc.jpg
0.428148	63002581105c4fc881100054f81f16a1	Drawing of an service half-cylinder and three half-disks.	2012.14304	fpixSTdrawing.jpg
0.492018	e04f8778e4ad47ecbdcfbfae3dceb65a	A picture of the robotic gantry as used for the assembly of the modules. Labels indicate the main tools used for the assembly.	2012.14304	gantryFull16labeled.jpg

0.502961

9b5cb981e9754fc5bd63fad1dd0edd8a

(a) Superimposed images of a mucosalivary spray emerging from the mouth of a 3D printed mannequin face obtained at 1000 fps over 4 seconds. Color bar indicates progression of time from white t=0 s (white) to t=4 s (orange). The mucosaliva is marked with a fluorescent tracer and deposits over time onto a horizontal surface. (b) The light intensity, I(x,y), corresponding to mucosaliva deposited on the surface in front of the head after n_s = 5 (top) and n_s =20 (bottom) expirations, or sprays (H = 0.5 m). Images are taken approximately 10 seconds after the last spray. A deposition lobe directed along the symmetry axis in front of the face can be observed. The laser line is set at y=0 m. (c) Images of deposited mucosalivary fluid after n_s = 20 sprays illuminated by a 1 cm wide laser light sheet at distance from source of (i) x = 0.35 m, (ii) x=0.5 m, and (iii) x=1.0 m.

2012.14268

Fig1_schematic.jpg

0.450382

d144b50667ba4f099c9fd09c8fb4438d

(a) Side view snapshot of a mucosalivary spray at t=0.35 s with a cloud threshold boundary (green line), where ∘ marks the cloud geometric center, and ▹ and ▿ mark the furthest x extent and the lowest z extent of the boundary, respectively. The green square and diamond indicates the furthest and lowest visible droplets in the x- and z-directions, respectively. (b) Tracked boundary of the cloud over time and the cone angle θ_s. The geometric center of the cloud asymptotes towards x=0.4 m, whereas individual droplets separate from the cloud travel further. The initial propagation speed is u_0=5 m/s. (d) The settling speed for the center, boundary, and individual droplet is found to be u_sc = 0.5 m/s, u_sb = 0.8 m/s, and u_sd = 2.0 m/s, respectively. Each point represents the mean over n_t = 10 trials with error bars of one standard deviation.

2012.14268

Fig2_cloudBoundary.jpg

0.506667

f1a737ec43944655b2bff1940354e298

Distribution profile of mucosalivary droplets deposited on a horizontal surface along x-axis from heights of (a) H=0.5 m, (b) H=1 m, and (c) H=1.5 m (n_s = 20). Insets are the respective semi-log plots. The noise floor is ϕ_noise=4.4×10^-6, which is three orders of magnitude less that the maximum signal. A total of n_t=12 trials are conducted for each height. Data points are binned with a bin size of 0.1 m. Distribution width σ_x ≈ 0.5 m for all heights. Black line is a fitted exponential function. (d) Maximum amount of material, ϕ_m decreases with height. (e) The mean of the distribution, x̅, decreases linearly with increasing height. (f) Distance x^(98%) at which 98% of the material is deposited decreases with H. Each point represents the mean over n_t=12 trials with error bars of one standard deviation.

2012.14268

Fig3_I-r_varyH.jpg

0.444621

75bfe178c59a4739abf480b1caaf4cbb

(a) The time evolution of the volume of droplets over a range of x=[0.7, 0.8] m. Droplets are expelled from a height of H=1.5 m. Volumes are self-normalized by the maximum volume such that ϕ/ϕ_∞. Time is normalized by the gravitational settling time, t_g=√(2H/g)∼ 0.55 s. Solid blue line is the average of n_t=5 trials (light gray lines). (b) The landing time, τ, of droplets at H=1.5 m. The time it takes to reach a percentage of ϕ_∞, τ, is plotted for x=40-70 cm. At a given x, the time to reach 25%, 50%, 75%, 90%, and 95% of ϕ_∞ goes from left to right. Error bars are one standard deviation from the mean.

2012.14268

Fig4_landingTime.jpg

0.478339

6d7b466cf71e4739a4372f309efb7b60

Trajectories T1-T4 of individual droplets corresponding to some of the brighter traces in Fig. <ref>(a). (a) Some droplets travel with pure projectile motion as shown by good agreement with Eq. (<ref>). (b) Trajectories T3 and T4 deviate from projectile motion due to air drag. The initial conditions are for T1: θ=-0.7^∘, u_0=2.25 m/s; T2: θ=25^∘, u_0=1.95 m/s; T3: θ=45^∘, u_0=1.8 m/s; and T4: θ=-20^∘, u_0=2.0 m/s.

2012.14268

Fig5_dropTrajectories.jpg

0.430411

93f996ab03f041f2b757382eee4d1ec2

(a) The average 2D projected area of the cloud, A_c, over time increases linearly after t=0.2 s (blue circles) with a slope of Ȧ_c=0.2 m^2/s. Error bars are one standard deviation. (b) The average settling speed of the cloud relative to the average effective diameter, D_c=√(4A_c/π) (blue circles) after t>0.2 s. The lines are implicit calculations of the terminal velocity as a function of discrete diameters. Our data falls within the the calculations in which cloud density, ρ_c, is between 0.5-2.5% greater than the density of air, ρ_a. An estimated density ratio of ρ_c/ρ_a= 1.011 agrees well with the experimental data. Gray circles indicate the the cloud growth for all n_t=10 trials. One standard deviation for diameter and velocity is shown in lower right corner.

2012.14268

Fig6_settlingSpeed.jpg

0.423309

10d769db810a4153ba6e578475ed5fc7

(a) Comparing the distribution profile of mucosalivary droplets for no mask vs with mask from a height of 1.5 m. (b) Same comparison as above on semilog scale. n_t=5 trials for both cases. (c) Switching to a N95 mask reduces the dispersal to within the noise floor of our measurements. Wearing a standard non-medical or N95 mask significantly reduces the total volume by at least 99%, and reduces the peaks recorded by at least 99.5%. Noise floor is ϕ_noise=4.4×10^-6, as shown in gray.

2012.14268

Fig7_mask-noMask.jpg

0.510738

bfce0fbd79b8414ea2e71865035fbcb9

Proposed manga translation framework. N' represents the translation of a source sentence N. Mitsuki Kuchitaka

2012.14271

context_translation.jpg

0.433214

6f29ff897b264f80ae8b244b49e67872

Results of fully automatic manga translation from Japanese to English and Chinese. Masami Taira, Syuji Takeya

2012.14271

e2e_example_multilang.jpg

0.410126

5ed16702f5644ed4a5e0f128349d3501

Given a manga page, our system automatically translates the texts on the page into English and replaces the original texts with the translated ones. Mitsuki Kuchitaka.

2012.14271

eg_first.jpg

0.435301

b270af186c7645f18b4d48a1bb607687

Outputs of the sentence-based (center) and frame-based (right) models. The values after H and B are respectively the human evaluation and BLEU scores for each page. Mitsuki Kuchitaka

2012.14271

eg_swapped.jpg

0.446362

bfc80347af3b402ba541859cbe570449

Translations output by the sentence-based model without (i) and with visual information (ii). By overwriting the character face in the input image (a) with a male face (b), the pronouns in the translation results (iii) are also changed. Nako Nameko

2012.14271

eg_visual.jpg

0.416672

281e7fb1a585463b9c45aca52138f14d

Translation results with and without visual features. Miki Ueda, Satoshi Arai.

2012.14271

eg_visual_2.jpg

0.447479

a77928a4d5f7455a976aaa51d64ac71d

Proposed framework of parallel corpus construction.

2012.14271

framework_collecting_groundtruth.jpg

0.448261

fc4a184944424b65b7bb4b79f1dc1cbf

GUI of the user study. Mitsuki Kuchitaka

2012.14271

gui_v2.jpg

0.464935

2812e6dc2d524edc95d507899cc94ce4

Procedures of rendering the training data.

2012.14271

synth_generation.jpg

0.411989

5dbed917ef944b4e8e575e61a8ed4fec

Term definition for manga text. Mitsuki Kuchitaka

2012.14271

term_definition.jpg

0.407008

8234aef7ad354add811d5f601fdc4fc2

Rules-based text-line detection.

2012.14271

text_detection_recognition_v4.jpg

0.419667

e80ede6d45f74664846e8f510ffacb4a

Generation of textline annotations.Yasuyuki Ohno

2012.14271

textline_detector.jpg

0.423891

5fb41be6219e42abaff7889f4841393b

Sample images with main upper-body line and clothing brightness value (bv). In the full dataset, these images are (from left to right) at brightnessvalue percentiles 0, 33, 66, and 100.

2012.14274

color_quantiles.jpg

0.441075

f4bcc5c04a724335869be897993f0cc0

2012.14274

photoshop_1.jpg

0.465732

d41d18d1e3fa489ca752e9d6f22b0735

2012.14274

photoshop_1_dark.jpg

0.448865

d91c63b78d014c92a1237be942d3340f

2012.14274

photoshop_1_light.jpg

0.430381

65490ef5844c4127a253fd43e42e6792

2012.14274

photoshop_2.jpg

0.494076

782ea4dfa61c469e9aac02e7229bbbe8

2012.14274

photoshop_2_dark.jpg

0.500136

6e8d04d92249427eb1499f9118886634

2012.14274

photoshop_2_light.jpg

0.468993

be34dd5549834cf5b6ef40b792fc9dca

BMI

2012.14274

plot_dist_bmi_pairwise_est_clean_dark_light.jpg

0.430929

0c934508b23940269eb0f2a6490124e2

BMI

2012.14274

plot_dist_bmi_pairwise_est_clean_dark_striped.jpg

0.463908

7e753dda0f0e471ba96756d74ad1aeb7

2012.14274

plot_dist_bmi_pairwise_est_clean_dl.jpg

0.434811

c56a27933cff4dc8a0dbd7c2d2f3dc6b

BMI

2012.14274

plot_dist_bmi_pairwise_est_clean_light_striped.jpg

0.403667

f3fcb96759fa48a095c6d43d3026a1d1

BMI

2012.14274

plot_dist_bmi_pairwise_est_dirty_dark_light.jpg

0.459425

be81486941ea4551a95c33323b82a548

BMI

2012.14274

plot_dist_bmi_pairwise_est_dirty_dark_striped.jpg

0.420217

69f2a23b42c8442e8a08278544a624ed

2012.14274

plot_dist_bmi_pairwise_est_dirty_dl.jpg

0.447394

e342276c8c644ea8b90c226079d4fd92

BMI

2012.14274

plot_dist_bmi_pairwise_est_dirty_light_striped.jpg

0.463984

712406bf61ba4654b05368887369fd01

2012.14274

plot_dist_bmi_pairwise_est_dirty_ls.jpg

0.434953

138ae42d07584ac28e1d7fbb5bb345ef

Height

2012.14274

plot_dist_h_pairwise_est_clean_dark_light.jpg

0.43767

910b67221eae44579fd24353bfe2ddbf

Height

2012.14274

plot_dist_h_pairwise_est_clean_dark_striped.jpg

0.4223

65ae1c03cebc4066bc613ae707aec937

2012.14274

plot_dist_h_pairwise_est_clean_dl.jpg

0.426563

68a15bc29a58419aa549b7be24546b1a

Height

2012.14274

plot_dist_h_pairwise_est_clean_light_striped.jpg

0.453957

6218852ea5fa4943a37d78527ad69499

Height

2012.14274

plot_dist_h_pairwise_est_dirty_dark_light.jpg

0.462967

f1e33ddbf2804b4a96db455901a75ebd

Height

2012.14274

plot_dist_h_pairwise_est_dirty_dark_striped.jpg

0.43535

ae8ec2b2143c4ea9b220b0872445b3e5

2012.14274

plot_dist_h_pairwise_est_dirty_dl.jpg

0.499694

03331079a5cb4fd6a34928514b9b3afd

Height

2012.14274

plot_dist_h_pairwise_est_dirty_light_striped.jpg

0.416175

7302fb6e37054ec4a2315d06dfe661c0

2012.14274

plot_dist_h_pairwise_est_dirty_ls.jpg

0.427711

5d511571e8954e31ab252067dbce4eb6

Weight

2012.14274

plot_dist_w_pairwise_est_clean_dark_light.jpg

0.52511

0beed24f527a4da890dc068599d7002e

Weight

2012.14274

plot_dist_w_pairwise_est_clean_dark_striped.jpg

0.476577

4f4fdfbb263c4a72b899fb7782f31775

2012.14274

plot_dist_w_pairwise_est_clean_dl.jpg

0.456272

098acac093f1402a8fb8c30717397acf

Weight

2012.14274

plot_dist_w_pairwise_est_clean_light_striped.jpg

0.467213

7f71c1e710dc41ce96ef099d78cf1bfb

Weight

2012.14274

plot_dist_w_pairwise_est_dirty_dark_light.jpg

0.454529

d1c8c6fffa6e4931b22d117439cb0741

Weight

2012.14274

plot_dist_w_pairwise_est_dirty_dark_striped.jpg

0.451635

f6f278f099ea4594983015709cd4b72d

2012.14274

plot_dist_w_pairwise_est_dirty_dl.jpg

0.442856

7709625cc7cf4e2bb5719a4249d50e7f

Weight

2012.14274

plot_dist_w_pairwise_est_dirty_light_striped.jpg

0.476604

2e1df82ee6b046649a957882ad80312b

2012.14274

plot_dist_w_pairwise_est_dirty_ls.jpg

0.452696

fccd29bd76844eb4948f482ed89f7abc

2012.14274

plot_scatter_bmi_pairwise_comp_clean_dl.jpg

0.463921

71925f45158f4ef49334176de9b78403

2012.14274

plot_scatter_bmi_pairwise_comp_dirty_dl.jpg

0.398461

bc0fe06e9209484593fdcd79c2d85122

2012.14274

plot_scatter_bmi_pairwise_comp_dirty_ls.jpg

0.454578

459e15247f3d4d7abf05e0a9a274f3c2

2012.14274

plot_scatter_h_pairwise_comp_clean_dl.jpg

0.436289

e7bfe34fa9064bd69cb63598a2caf3de

2012.14274

plot_scatter_h_pairwise_comp_dirty_dl.jpg

0.439047

bf16322260c54e54a103447f64a5ed9d

2012.14274

plot_scatter_h_pairwise_comp_dirty_ls.jpg

0.462887

e31c575f11994a10b34eb7d58031c971

2012.14274

plot_scatter_w_pairwise_comp_clean_dl.jpg

0.416554

cbd340d9274c40e9affffe112038ab79

2012.14274

plot_scatter_w_pairwise_comp_dirty_dl.jpg

0.440359

3b862f22de2941fab86f9a15f13a20f5

2012.14274

plot_scatter_w_pairwise_comp_dirty_ls.jpg

0.480596

b2dd42aadcd14cdc8bf035e623ef6ec1

Brightness vector

2012.14274

plot_stripes_colorline.jpg

0.479454

f319087f5eb84e279086253e5f356ac0

Fourier spectrum

2012.14274

plot_stripes_colorline_fourier.jpg

0.426065

e1e9401b8b5e403a9e786b4f5d753510

Smoothed brightness vector

2012.14274

plot_stripes_colorline_smooth.jpg

0.460572

2c0081b28460413590b067dda6ddce9d

Image

2012.14274

plot_stripes_sample.jpg

0.378165

5408f317bf6748df87d0cfbe8db11c56

Schematic of direction discovery and image transformation. The method is broadly useful with different generators and could be augmented with an inverter to take real-world images as inputs.

2012.14283

z_L1_schematic_updated.jpg

0.437045

74ffe99ae51840af8a9295044901cb91

Learning by ignoring. Along the solid arrows, predictions are made and train/validation losses are calculated. Along the dotted arrows, gradients are calculated and parameters (including ignoring variables and network weights) are updated.

2012.14288

archi-2-lbi.jpg

0.432753

f199dca50dfb482f80960eac6a8b1b31

Illustration of learning by ignoring. Void circles denote ignored pretraining data examples. Given a set of intermediately selected pretraining examples, they are used to pretrain a model, which is then finetuned and validated. The validation performance provides guidance on what pretraining examples should be ignored in the next round of learning. This process iterates until convergence.

2012.14288

illustration-lbi.jpg

0.518656

69ad135e33fe486d9e03c66b66ad6695

This irresistable Far Side cartoon by the magnificent Gary Larson does not quite express the premise of this BoF. To illustrate that the Users, Developers and Managers of scientific software are holding each other back by the way they are organized, the formation of the groundbound waddlers should of course be different from the lofty flyers. But we trust that you get the point anyway.

2012.14290

B11-143_f1.jpg

0.453781

e6ac24932fe24bc19e53666063d28114

The evolutionary behaviour of the deceleration parameter q versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig10.jpg

0.53439

3ac59a068ce546aa9da402ed5a44be12

The evolutionary behaviour of the Barrow energy density parameter Ω_D versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig11.jpg

0.438543

b54096c049ea4fb5a28a60cca43bbd43

The evolutionary behaviour the squared sound speed v_s^2 versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig12.jpg

0.466106

086f8ed482c44a05a093de672a2ed21f

The evolutionary behaviour of equation of state parameter ω_D versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig21.jpg

0.491952

f095dec1252b4110b0a50b353e976deb

The evolutionary behaviour of the deceleration parameter q versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig22.jpg

0.462113

73159a5b7e7641e1a643b3e80f365256

The evolutionary behaviour of the Barrow energy density parameter Ω_D versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig23.jpg

0.431694

877873fe8f214d17936109a0892934cb

The evolutionary behaviour of the squared sound speed v^2_s versus redshift for interacting NBADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig24.jpg

0.506608

c36264b7ecfb47cba5b70203d9c0d419

The evolutionary behaviour of equation of state parameter ω_D versus redshift for interacting BADE. Here, we have considered H_0=67, Ω_D0 =0.70, Δ = - 5 and C=10.

2012.14291

fig9.jpg

0.505152

dc231e7a3fda467797c7e289032fc115

Stability potential (<ref>) for a symmetric kink with power-law asymptotics: the potential is V(φ) = 1/2cos^4φ, the corresponding kink is φ_ K^(x) = arctg x.

2012.14298

fig1.jpg

0.457744

c8eb67f114444911ad784943e44bbab5

The spectrum according to eq. p5_1 for a string with N=40 sections, x=0.3 and L=1. The dots denote the eigenvalues. For N→∞ or L→∞ these approach the solid line.

2012.14301

fig1_v2.jpg

0.524014

7f031df3a7754a61a3a9da2219e762bf

The vacuum energy E_0, E01, E2, for a composite string as a function of the ratio x, 2_3, of the string tensions for several numbers N of sections of the string. N increases from top to bottom (color online). The length of the string is L=1.

2012.14301

fig2_v2.jpg

0.428764

3d13a6f45cb149a7a650b7a1bc1f76bc

The limiting slope w(x) of the vacuum energy4.6 for N→∞ as a function of the ratio x.

2012.14301

fig3_v2.jpg

0.635904

14cbb49478ca434ebd23f8bbf806c799

Drawing of one detector half in the rϕ view, showing the arrangement of ladders in the four detector layers.

2012.14304

BPIX-Sensor-Module-Positions2.jpg

0.430248

f026a15fd8ac41128aff54e8ee276690

One half of the detector.

2012.14304

BPIXFinal.jpg

0.398587

76be6be848234491bcdd7c9d4e016b69

Picture of a service half-cylinder during construction. The service half-cylinder is placed on a rotatable mandrel. Segments A and B are made from Airex foam, while Segment C is made from carbon fiber plates.

2012.14304

BPIXSTUZH.jpg

0.404951

ae01aae6a0ee42fcb711960e0d3928cd

Thermal image of an disk during quality control. An infrared camera observed the disk through a porthole in the test enclosure and was programmed such that the allowed temperature range, between -20.0 and -10.5 , was rendered in the color spectrum from purple to white, while temperatures outside of the allowed range were rendered green. The blade surfaces of the disk, except for the areas where the heaters and co-cured reflective pads are located (corresponding to the green large and small rectangles), appear to be within the acceptable range.

2012.14304

FPIX_disk_thermal.jpg

0.395259

e691d7bc995e4e449f89e9840d7d836c

Cluster hit efficiency as a function of instantaneous luminosity measured using tracks from collisions in 2018.

2012.14304

Hitefficiency_new.jpg

0.444936

ebc7f37af36a4a13be02753dd861c53f

Single-pixel hit efficiency versus X-ray hit rate measured for double columns in a ROC of a L2 module. The dashed lines indicate the values at which the single-pixel hit efficiency per double column is determined during the module qualification and cover the range of expected hit rates in L2-L4. The solid red line is a fit to the data points using a polynomial function.

2012.14304

L2efficiency.jpg

0.416283

26c24bc5270c490eaff153db2bceee48

Picture of the L2 mechanics during production.

2012.14304

L2mechanics.jpg

0.454154

854006f43ca9410baccf420979979e01

Layout of the compared to the original detector layout, in longitudinal view.

2012.14304

Layout.jpg

0.434217

627d2b1369f44ffd928f6bfb2527b132

Drawings of the pixel detector modules for L1 (left), L2-4 (middle), and the detector (right).

2012.14304

Modules.jpg

0.471011

22b04715db8d4f40aa6bd4d7deb40ec5

Pulse height measured in one pixel as a function of amplitude of the internal calibration signal in units of . The slope of the curve below saturation is fitted with a linear function.

2012.14304

PHvsVcal.jpg

0.413213

edd7deae3d124664a0f1c3d823142038

DC photocurrent measured at the FED receiver as a function of the setting of the laser bias current on the POH. The measured values for one FED (24 optical links) are shown. Each curve corresponds to one optical link. The bias current is programmable through the interface in steps of 0.45 mA per setting. The setting is adjusted such that the output current is above 0.1 mA (indicated by the horizontal line). The spread between optical links is caused by differences in the quality of the optical connection.

2012.14304

POHBias.jpg

0.456981

2914af05ec144a23a46a697b18372b0d

Drawing of the Phase-1 and detectors together with the service half-cylinders that hold the readout and control circuits as well as power and cooling lines.

2012.14304

PixelSystem.jpg

0.462632

226942498b3545959c7d334ebf13acc5

Picture of one side of the CMS detector with the endcaps opened and the temporary platform for the pixel detector installation set up. Pictures used by permission from CERN.

2012.14304

SSP_99.jpg

0.580188

fe7b2d04d58e47ce931565c5f9a070a5

Calibration of one ROC on a pixel detector module with monochromatic X-rays. The measured pulse height is plotted against the expected number of electrons from the K_α-transitions in the different target materials, as given in Ref. <cit.>. The pulse height in units of the internal calibration signal has a linear dependence on the number of electrons expected to be collected from different target materials. The red line shows a fit with a linear function to the data points.

2012.14304

XrayCalibrationChipResults.jpg

0.409497

697b5fd344b246dd9175b7c5586937f3

Drawing of a service half-cylinder.

2012.14304

bpixSTDrawingNew.jpg

0.455869

1f0d55c4b82944d8925192dc0a2c667a

Measured single-pixel hit efficiency for the as a function of pixel hit rate in the 200 MeV proton beam at PSI. The version of the used in the CMS experiment in operation in 2017 and 2018 is compared to the revised version of the chip that will be used in the replacement of L1 in Run 3. The yellow and green points labeled "with resets" correspond to the data-taking mode in which every trigger signal was preceded by a reset signal.

2012.14304

effproc.jpg

0.428148

63002581105c4fc881100054f81f16a1

Drawing of an service half-cylinder and three half-disks.

2012.14304

fpixSTdrawing.jpg

0.492018

e04f8778e4ad47ecbdcfbfae3dceb65a

A picture of the robotic gantry as used for the assembly of the modules. Labels indicate the main tools used for the assembly.

2012.14304

gantryFull16labeled.jpg