%PDF-1.3 1 0 obj << /Kids [ 3 0 R 4 0 R 5 0 R 6 0 R 7 0 R 8 0 R 9 0 R 10 0 R 11 0 R ] /Type /Pages /Count 9 >> endobj 2 0 obj << /Title (Categorizing Concepts With Basic Level for Vision\055to\055Language) /Producer (PyPDF2) /Author (Hanzhang Wang\054 Hanli Wang\054 Kaisheng Xu) /Subject (2018 IEEE Conference on Computer Vision and Pattern Recognition) >> endobj 3 0 obj << /Parent 1 0 R /Rotate 0 /Contents 13 0 R /Resources << /XObject << /x8 14 0 R /x6 17 0 R /x12 20 0 R /x10 23 0 R >> /ExtGState << /s9 26 0 R /s11 29 0 R /a0 << /CA 1 /ca 1 >> /R34 32 0 R /s5 35 0 R /R23 38 0 R /s7 39 0 R >> /Font << /F2 42 0 R /R28 43 0 R /F1 48 0 R /R32 49 0 R /R24 54 0 R /R30 57 0 R /R26 60 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 63 0 R 64 0 R 65 0 R 66 0 R 67 0 R 68 0 R 69 0 R 70 0 R 71 0 R 72 0 R 73 0 R 74 0 R ] /Type /Page >> endobj 4 0 obj << /Parent 1 0 R /Rotate 0 /Contents 75 0 R /Resources << /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /F2 76 0 R /F1 77 0 R /R24 54 0 R /R30 57 0 R /R26 60 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 78 0 R 79 0 R 80 0 R 81 0 R 82 0 R 83 0 R 84 0 R 85 0 R 86 0 R 87 0 R 88 0 R 89 0 R 90 0 R 91 0 R 92 0 R 93 0 R 94 0 R 95 0 R 96 0 R 97 0 R 98 0 R 99 0 R 100 0 R 101 0 R 102 0 R 103 0 R 104 0 R 105 0 R ] /Type /Page >> endobj 5 0 obj << /Parent 1 0 R /Rotate 0 /Contents 106 0 R /Resources << /XObject << /R91 107 0 R >> /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /R94 108 0 R /F2 113 0 R /R89 114 0 R /F1 119 0 R /R30 57 0 R /R92 120 0 R /R87 125 0 R /R24 54 0 R /R85 130 0 R /R26 60 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 135 0 R 136 0 R 137 0 R 138 0 R 139 0 R 140 0 R 141 0 R 142 0 R 143 0 R 144 0 R 145 0 R 146 0 R ] /Type /Page >> endobj 6 0 obj << /Parent 1 0 R /Rotate 0 /Contents 147 0 R /Resources << /XObject << /R116 148 0 R /R117 149 0 R /R114 150 0 R /R115 151 0 R >> /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /R108 152 0 R /R112 157 0 R /R110 162 0 R /R106 167 0 R /F2 172 0 R /R89 114 0 R /F1 173 0 R /R87 125 0 R /R24 54 0 R /R85 130 0 R /R26 60 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 174 0 R 175 0 R 176 0 R ] /Type /Page >> endobj 7 0 obj << /Parent 1 0 R /Rotate 0 /Contents 177 0 R /Resources << /XObject << /R147 178 0 R >> /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /R112 157 0 R /R148 179 0 R /R106 167 0 R /F2 184 0 R /R89 114 0 R /F1 185 0 R /R30 57 0 R /R24 54 0 R /R85 130 0 R /R26 60 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 186 0 R 187 0 R 188 0 R 189 0 R 190 0 R 191 0 R 192 0 R 193 0 R 194 0 R 195 0 R 196 0 R 197 0 R 198 0 R 199 0 R 200 0 R 201 0 R 202 0 R 203 0 R 204 0 R 205 0 R 206 0 R 207 0 R 208 0 R ] /Type /Page >> endobj 8 0 obj << /Parent 1 0 R /Rotate 0 /Contents 209 0 R /Resources << /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /F2 210 0 R /F1 211 0 R /R24 54 0 R /R26 60 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 212 0 R 213 0 R 214 0 R 215 0 R 216 0 R 217 0 R 218 0 R 219 0 R 220 0 R 221 0 R 222 0 R 223 0 R 224 0 R 225 0 R 226 0 R 227 0 R 228 0 R 229 0 R 230 0 R 231 0 R 232 0 R 233 0 R 234 0 R 235 0 R 236 0 R 237 0 R 238 0 R 239 0 R 240 0 R 241 0 R 242 0 R 243 0 R 244 0 R 245 0 R 246 0 R 247 0 R 248 0 R 249 0 R 250 0 R 251 0 R 252 0 R 253 0 R 254 0 R 255 0 R 256 0 R 257 0 R ] /Type /Page >> endobj 9 0 obj << /Parent 1 0 R /Rotate 0 /Contents 258 0 R /Resources << /XObject << /R213 259 0 R /R215 260 0 R /R214 261 0 R /R217 262 0 R /R216 263 0 R /R218 264 0 R >> /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /F2 265 0 R /F1 266 0 R /R24 54 0 R /R30 57 0 R /R26 60 0 R >> /ProcSet [ /Text /ImageC /ImageB /PDF /ImageI ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 267 0 R 268 0 R 269 0 R 270 0 R 271 0 R 272 0 R 273 0 R ] /Type /Page >> endobj 10 0 obj << /Parent 1 0 R /Rotate 0 /Contents 274 0 R /Resources << /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /F2 275 0 R /F1 276 0 R /R24 54 0 R /R30 57 0 R /R26 60 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 277 0 R 278 0 R 279 0 R 280 0 R 281 0 R 282 0 R 283 0 R 284 0 R 285 0 R 286 0 R 287 0 R 288 0 R 289 0 R 290 0 R 291 0 R 292 0 R 293 0 R 294 0 R 295 0 R 296 0 R 297 0 R 298 0 R 299 0 R 300 0 R 301 0 R 302 0 R 303 0 R 304 0 R 305 0 R 306 0 R 307 0 R 308 0 R 309 0 R 310 0 R ] /Type /Page >> endobj 11 0 obj << /Parent 1 0 R /Rotate 0 /Contents 311 0 R /Resources << /ExtGState << /R34 32 0 R /R23 38 0 R >> /Font << /F2 312 0 R /F1 313 0 R /R30 57 0 R /R26 60 0 R >> /ProcSet [ /ImageC /Text /PDF /ImageI /ImageB ] >> /CropBox [ 0 0 612 792 ] /MediaBox [ 0 0 612 792 ] /Annots [ 314 0 R 315 0 R 316 0 R 317 0 R 318 0 R 319 0 R 320 0 R 321 0 R 322 0 R 323 0 R 324 0 R 325 0 R 326 0 R 327 0 R 328 0 R 329 0 R 330 0 R 331 0 R 332 0 R 333 0 R 334 0 R 335 0 R 336 0 R 337 0 R 338 0 R 339 0 R 340 0 R 341 0 R 342 0 R 343 0 R 344 0 R 345 0 R 346 0 R 347 0 R 348 0 R 349 0 R 350 0 R 351 0 R 352 0 R 353 0 R 354 0 R 355 0 R 356 0 R 357 0 R 358 0 R 359 0 R 360 0 R 361 0 R 362 0 R 363 0 R 364 0 R 365 0 R 366 0 R 367 0 R 368 0 R 369 0 R 370 0 R 371 0 R 372 0 R 373 0 R ] /Type /Page >> endobj 12 0 obj << /Type /Catalog /Pages 1 0 R >> endobj 13 0 obj << /Length 21838 >> stream q q q 0.1 0 0 0.1 0 0 cm /R23 gs 0 0 0 1 k q 10 0 0 10 0 0 cm BT /R24 14.3462 Tf 1 0 0 1 103.383 675.003 Tm [ (Categorizing) -249.691 (C) -0.0987 (oncepts) -249.602 (with) -250.095 (Basic) -249.809 (L) 0.0936 (e) 14.9059 (v) 9.99625 (el) -250.192 (f) 24.9923 (or) -249.809 (V) 36.9048 (ision\055to\055Language) ] TJ /R26 11.9551 Tf 67.8598 -37.8543 Td [ (Hanzhang) -249.799 (W) 79.7969 (a) -0.2083 (ng) -1999.31 (Hanli) -250.5 (W) 79.7969 (ang) ] TJ /R28 7.9701 Tf 163.089 4.33828 Td <02> Tj /R26 11.9551 Tf 26.9012 -4.33828 Td [ (Kaisheng) -249.785 (Xu) ] TJ -0.0012 Tc -282.86 -13.95 Td [ (Department) -250.584 (of) -250.485 (Computer) -250.49 (Science) -250.21 (a) -0.30836 (nd) -249.895 (T) 69.5841 (echnology) 64.7973 (\054) -249.985 (T) 79.3864 (ongji) -250.59 (Uni) 25.0081 (v) 15.1098 (e) -0.30836 (rsity) 64.8096 (\054) -250.704 (S) -0.41251 (hanghai\054) -250.012 (P) 110.19 (\056) -249.199 (R\056) -250 (China) ] TJ 13.6086 -13.941 Td [ (K) 24.7932 (e) 14.7882 (y) -249.911 (L) -0.40435 (aboratory) -250.611 (of) -249.814 (E) -0.40435 (mbedded) -250.59 (S) -0.41251 (ystem) -249.691 (a) -0.30836 (nd) -249.895 (Service) -250.216 (C) -0.11436 (omputing\054) -251.511 (Ministry) -250.616 (of) -250.486 (Education\054) ] TJ 108.936 -13.9504 Td [ (T) 79.3864 (ongji) -249.806 (Uni) 25.0081 (v) 15.1098 (e) -0.30836 (rsity) 64.8096 (\054) -250.704 (S) -0.41251 (hanghai\054) -250.012 (P) 110.19 (\056) -249.985 (R\056) -250.002 (China) ] TJ 0.0012 Tc /R24 11.9551 Tf -54.9184 -41.0492 Td (Abstract) Tj -0.001 Tc /R30 9.9626 Tf -83.9352 -29.9781 Td [ (V) 74.2836 (i) -0.28917 (sion\055to\055langua) 10.3059 (g) 10.2949 (e) -320.404 (tasks) -317.486 (r) 36.6054 (equir) 36.6103 (e) -318.6 (a) -317.69 (uni\336ed) -319.505 (semantic) -317.702 (un\055) ] TJ -0.00199 Tc -11.9523 -11.9523 Td [ (der) 9.40529 (standing) -414.387 (of) -414.202 (visual) -414.203 (content\056) -800.816 (Howe) 15.6077 (ver) 110.594 (\054) -456.593 (t) -0.38964 (he) -413.507 (information) ] TJ -0.001 Tc 11.952 TL T* [ (contained) -464.008 (in) -462.189 (ima) 10.301 (g) 10.2949 (e) 0.28794 (\057video) -463.997 (is) -462.109 (essentially) -462.2 (ambiguous) -464.815 (on) -463.094 (t\055) ] TJ -0.00199 Tc 11.9609 TL T* [ (wo) -402.707 (per) 9.40529 (spectives) -404.385 (manifested) -402.68 (on) -402.715 (the) -402.682 (d) 0.19727 (iver) 9.39794 (se) -402.685 (under) 9.40774 (standing) ] TJ -0.00299 Tc 11.9523 TL T* [ (among) -513 (dif) 17.5087 (f) -0.48889 (er) 36.4155 (ent) -511.905 (per) 9.30114 (sons) -512.896 (and) -512.093 (the) -511.199 (various) -512.887 (under) 9.29992 (standing) ] TJ -0.00199 Tc T* [ (gr) 14.7849 (ains) -245.403 (e) 15.5967 (ven) -245.504 (for) -244.502 (t) -0.38842 (he) -244.594 (same) -244.593 (per) 9.40529 (son\056) -308.495 (Inspir) 36.5062 (ed) -244.594 (by) -244.594 (the) -244.592 (basic) -244.587 (le) 15.6003 (vel) ] TJ 11.9613 TL T* [ (in) -283.399 (early) -284.316 (co) 10.187 (gnition\054) -294.017 (a) -283.402 (B) 0.11763 (asic) -284.308 (Concept) -285.914 (\050BaC\051) -284.193 (cate) 39.9811 (gory) -285.212 (i) -0.38964 (s) -284.185 (p) 0.19605 (r) 44.5832 (o) 0.19605 (\055) ] TJ 11.952 TL T* [ (posed) -301.482 (in) -300.612 (this) -300.489 (work) -300.613 (that) -301.314 (contains) -302.298 (both) -300.61 (consensus) -302.307 (and) -301.516 (pr) 44.5832 (op\055) ] TJ T* [ (er) -299.603 (le) 15.6003 (vel) -300.395 (of) -299.514 (visual) -299.516 (content) -301.29 (to) -298.809 (help) -299.717 (neur) 14.8174 (al) -301.317 (network) -299.715 (tac) 20.1094 (k) 0.18869 (le) -299.68 (the) ] TJ 11.9613 TL T* [ (abo) 10.1974 (ve) -294.281 (pr) 44.5839 (oblems\056) -437.695 (Speci\336cally) 55.3045 (\054) -305.793 (a) -292.499 (salient) -293.083 (concept) -294.912 (cate) 39.9811 (gory) -294.309 (i) -0.38964 (s) ] TJ 11.952 TL T* [ (\336r) 9.41326 (stly) -256.31 (g) 10.1956 (ener) 14.8076 (ated) -259.881 (by) -258.082 (inter) 9.41142 (s) -0.58569 (ecting) -258.083 (t) -0.38842 (he) -258.082 (labels) -257.994 (o) 0.19605 (f) -257.915 (I) -0.51707 (ma) 10.1944 (g) 10.1944 (e) 0.18869 (Net) -259.716 (and) ] TJ T* [ (the) -372.883 (vocab) 20.1008 (ulary) -373.782 (o) 0.19727 (f) -372.601 (M) -0.08087 (SCOCO) -373.999 (dataset\056) -677.1 (Then\054) -404.205 (accor) 36.4915 (ding) -374.714 (to) ] TJ 11.9609 TL T* [ (the) -363.787 (observation) -364.684 (f) -0.38842 (r) 44.5832 (o) 0.19727 (m) -364.097 (human) -364.681 (early) -363.792 (co) 10.187 (gnition) -364.709 (t) -0.38964 (hat) -364.483 (c) 15.5979 (h) 0.19605 (ildr) 36.5135 (e) 0.18869 (n) ] TJ -0.001 Tc 11.9523 TL T* [ (mak) 10.2899 (e) -296.016 (fe) 15.6824 (wer) -296.784 (m) 0.09925 (istak) 10.3034 (e) 0.28794 (s) -295.888 (o) 0.29652 (n) -296.007 (t) -0.28917 (he) -295.993 (basic) -295.099 (le) 15.6824 (vel\054) -308.387 (t) -0.28917 (he) -295.09 (salient) -296.617 (cate) 40.1061 (go\055) ] TJ -0.00199 Tc 11.9609 TL T* [ (ry) -349.399 (is) -348.407 (further) -350.212 (r) 36.5062 (e\336ned) -350.308 (by) -350.302 (clustering) -349.393 (concepts) -351.096 (w) 0.08822 (ith) -349.385 (a) -349.393 (de\336ned) ] TJ 11.952 TL T* [ (confusion) -361.111 (d) 0.19727 (e) 39.9847 (g) 0.19727 (r) 36.5062 (e) 0.18747 (e) -360.184 (w) 0.08822 (hic) 15.6016 (h) -361.115 (measur) 36.5013 (es) -360.103 (the) -360.218 (d) 0.19605 (if) 17.6123 (\336culty) -360.183 (for) -359.188 (con) 39.986 (vo\055) ] TJ -0.001 Tc T* [ (lutional) -303.014 (neur) 14.9032 (al) -304.803 (network) -303.21 (to) -302.294 (distinguish) -302.317 (class) -303.096 (pair) 9.50332 (s) -0.48521 (\056) -466.491 (F) 45.387 (inally) 55.4172 (\054) ] TJ -0.00199 Tc 11.9609 TL T* [ (a) -276.188 (p) 0.19727 (r) 36.5056 (e) 0.18747 (\055tr) 14.8192 (a) 0.19727 (ined) -277.995 (model) -277.8 (based) -277.995 (on) -276.187 (Goo) 10.1956 (gLeNet) -279.593 (is) -277.008 (pr) 44.5832 (oduced) -278.907 (with) ] TJ 11.9523 TL T* [ (the) -383.705 (p) 0.19727 (r) 44.5826 (oposed) -385.507 (BaC) -383.809 (cate) 39.9811 (gory) -384.606 (o) 0.19727 (f) -384.403 (1) 0.19727 (\054372) -383.699 (concept) -386.19 (classes\056) -710.516 (T) 92.3865 (o) ] TJ T* [ (verify) -228.292 (the) -229.182 (e) 0.18747 (f) 17.608 (f) -0.38903 (ectiveness) -229.991 (o) 0.19727 (f) -229.017 (t) -0.38842 (he) -228.283 (pr) 44.5845 (oposed) -230.082 (cate) 39.9811 (gorizing) -229.202 (method) ] TJ -0.001 Tc 11.9609 TL T* [ (for) -329.286 (v) 0.28794 (ision\055to\055langua) 10.2924 (g) 10.2949 (e) -330.284 (tasks\054) -348.088 (two) -329.396 (k) 0.28794 (inds) -329.303 (of) -329.193 (e) 20.205 (xperiments) -330.212 (a) 0.29652 (r) 36.6066 (e) ] TJ -0.00199 Tc 11.9523 TL T* [ (performed) -299.683 (including) -299.709 (ima) 10.1981 (g) 10.1956 (e) -298.782 (captioning) -300.609 (and) -299.712 (visual) -299.516 (question) ] TJ T* [ (answering) -199.405 (with) -198.499 (the) -198.48 (benc) 15.5905 (hmark) -202.089 (datasets) -199.295 (o) 0.19605 (f) -199.219 (M) -0.08087 (SCOCO\054) -200.106 (Flic) 20.1118 (k\055) ] TJ 11.9613 TL T* [ (r30k) -238.317 (and) -239.212 (COCO\055QA\056) -238.895 (The) -239.211 (e) 20.1057 (xperimental) -239.815 (r) 36.5062 (e) 0.18869 (sults) -238.105 (demonstr) 14.7843 (ate) ] TJ 11.952 TL T* [ (that) -230.816 (the) -230.985 (r) 36.5056 (epr) 36.4988 (esentations) -231.803 (derived) -230.998 (fr) 44.5857 (om) -230.393 (the) -230.986 (c) 0.18869 (o) 10.1944 (gnition\055inspir) 36.4841 (ed) ] TJ T* [ (BaC) -445.288 (cate) 39.9811 (gory) -446.008 (p) 0.19727 (r) 44.5832 (o) 0.19727 (mote) -446.009 (r) 36.5062 (epr) 36.4988 (esentation) -446.004 (l) -0.38964 (earning) -445.1 (of) -444.901 (neur) 14.8162 (al) ] TJ -0.001 Tc 11.9609 TL T* [ (networks) -364.509 (o) 0.29713 (n) -363.681 (v) 0.28794 (ision\055to\055langua) 10.2924 (g) 10.2949 (e) -365.494 (tasks\054) -391.492 (and) -364.616 (a) -362.819 (performance) ] TJ -0.00199 Tc 11.952 TL T* [ (impr) 44.5863 (o) 10.1956 (vement) -262.409 (is) -258.893 (gained) -260.781 (without) -260.601 (modifying) -260.814 (s) -0.58569 (tandar) 36.5135 (d) -259.917 (models\056) ] TJ ET Q 4.05 w 0 0 0 1 K 501.121 1471.49 m 1446.12 1471.49 l S q 10 0 0 10 0 0 cm BT /R32 5.9776 Tf 1 0 0 1 60.1379 140.538 Tm <02> Tj 0.0008 Tc /R26 7.9701 Tf 4.31992 -2.8082 Td [ (Corresponding) -294.918 (author) -295.878 (\050H\056) -296.488 (W) 80.1779 (ang\054) -306.584 (E\055mail\072) -402.792 (hanliw) 10.7059 (ang\100tongji\056edu\056cn\051\056) ] TJ 0 Tc -14.3461 -9.46758 Td [ (This) -290.816 (w) 10.503 (o) -0.30709 (rk) -290.486 (w) 10.601 (a) 0.20294 (s) -290.787 (s) 0.48705 (upported) -290.476 (in) -290.496 (part) -289.982 (by) -290.516 (National) -291.103 (N) 0.40587 (atural) -290.015 (S) 0.40741 (cience) -292.282 (F) 15.1108 (oundation) ] TJ 0.0008 Tc 9.46758 TL T* [ (of) -256.419 (China) -254.905 (under) -255.208 (G) 0.50696 (rants) -256.793 (61622115) -254.214 (and) -256.49 (61472281\054) -255.788 (Program) -255.214 (for) -256.4 (P) 0.50849 (rofessor) -256.384 (o) -0.2083 (f) ] TJ 0 Tc 9.45898 TL T* [ (Special) -269.721 (Appointment) -267.384 (\050Eastern) -269.008 (S) 0.40741 (cholar\051) -268.896 (a) 0.20217 (t) -268.623 (S) 0.40741 (hanghai) -267.39 (I) -0.07964 (nstitutions) -269.385 (of) -267.795 (Higher) ] TJ 0.0008 Tc 9.46797 TL T* [ (Learning) -382.984 (\050No\056) -707.508 (GZ2015005\051\054) -415 (S) 0.50849 (hanghai) -382.503 (E) 0.22361 (ngineering) -381.798 (R) -0.28794 (esearch) -384.111 (Center) -381.706 (of) ] TJ 0 Tc 9.45859 TL T* [ (Industrial) -190.584 (V) 60.2004 (i) 0.2037 (sion) -191.117 (Perception) -189.989 (\046) -190.905 (Intelligent) -189.498 (C) -0.3875 (omputing) -190.023 (\050) -0.07964 (17DZ2251600\051\054) -199.399 (a) 0.20217 (nd) ] TJ 9.46797 TL T* [ (IBM) -249.314 (S) 0.40817 (hared) -250.938 (U) 0.40664 (ni) 24.9896 (v) 14.3955 (e) 0.20217 (rsity) -251.019 (Research) -251.004 (A) 91.91 (w) 10.5022 (ards) -250.195 (Program\056) ] TJ /R24 11.9551 Tf 258.75 473.318 Td [ (1\056) -249.899 (Intr) 17.9035 (oduction) ] TJ -0.00199 Tc /R26 9.9626 Tf 11.952 -18.9629 Td [ (F) 15.615 (o) 0.19605 (r) -313.092 (human) -314.102 (being\054) -328.418 (basic) -312.311 (le) 25.481 (v) 15.6052 (e) 0.18624 (l) -313.904 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 444.699 535.296 Tm (39) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.001 Tc /R26 9.9626 Tf 1 0 0 1 454.662 535.296 Tm [ (\135) -312.09 (i) -0.28917 (s) -311.218 (a) 0.28917 (n) -312.198 (abstraction) -313.1 (t) -0.28917 (hat) ] TJ -145.8 -11.9523 Td [ (plays) -284.982 (a) -283.313 (p) 0.29652 (rimary) -283.284 (role) -284.19 (in) -283.318 (the) -283.306 (f) -0.49501 (orm) -283.61 (o) 0.29897 (f) -284.095 (early) -283.315 (cognition\056) -413.201 (It) -283.09 (con\055) ] TJ 11.952 TL T* [ (tains) -220.01 (s) -0.48521 (e) 25.6158 (v) 15.7057 (eral) -221.701 (objects) -220.917 (t) -0.28917 (hat) -220.804 (are) -219.218 (\336) 0.30877 (rstly) -219.184 (percei) 25 (v) 15.7082 (e) 0.28917 (d) -222.801 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 491.048 511.392 Tm (39) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.00199 Tc /R26 9.9626 Tf 1 0 0 1 501.002 511.392 Tm [ (\135) -220.086 (and) -221.105 (f) 9.40529 (a) 0.18624 (ster) ] TJ -192.141 -11.9609 Td [ (being) -343.082 (named) -343.087 (by) -342.097 (children\054) -365.405 (then) -342.105 (other) -342.894 (subordinate) -343.011 (concept\055) ] TJ -0.001 Tc 11.9523 TL T* [ (s) -323.882 (a) 0.28672 (re) -324.899 (learned) -324.919 (afterw) 9.98118 (ards) -325.715 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 412.405 487.479 Tm (45) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.001 Tc /R26 9.9626 Tf 1 0 0 1 422.359 487.479 Tm [ (\135\056) -532.387 (In) -323.985 (the) -324.005 (t) -0.28917 (axonomic) -326.715 (hierarch) 5.70493 (y) 65.3861 (\054) ] TJ -113.499 -11.952 Td [ (basic) -313.105 (le) 25.6036 (v) 15.7057 (e) 0.28672 (l) -313.805 (i) -0.28917 (s) -312.982 (l) -0.28917 (ocated) -313.092 (on) -313.119 (the) -313.105 (m) 0.10783 (iddle) -313.095 (w) 0.09802 (hich) -313.119 (represents) -313.918 (m) 0.10783 (id\055) ] TJ 0 Tc 11.9613 TL T* [ (le) 25.6844 (v) 15.8062 (e) 0.38719 (l) -337.191 (generality) -336.512 (\050) ] TJ -0.15741 Tc /R30 9.9626 Tf 68.8148 0 Td [ (e\056) -15.9973 (g) ] TJ -0.001 Tc /R26 9.9626 Tf 11.7359 0 Td [ (\056\054) -357.195 (bird\054) -357.21 (chair\051\054) -357.19 (oppose) -336.601 (t) -0.28917 (o) -334.782 (m) 0.10783 (ore) -335.699 (general) ] TJ -0.00299 Tc -80.5508 -11.952 Td [ (concepts) -259.91 (\050) ] TJ -0.15741 Tc /R30 9.9626 Tf 40.743 0 Td [ (e\056) -15.9973 (g) ] TJ -0.001 Tc /R26 9.9626 Tf 11.7359 0 Td [ (\056\054) -260.506 (v) 15.7057 (e) 0.28672 (rtebrate\054) -261.402 (furniture\051) -257.888 (and) -258.016 (more) -257.996 (speci\336c) -258.913 (con\055) ] TJ -0.00299 Tc -52.4789 -11.9523 Td [ (cepts) -342.1 (\050) ] TJ -0.15741 Tc /R30 9.9626 Tf 27.1891 0 Td [ (e\056) -15.9973 (g) ] TJ -0.001 Tc /R26 9.9626 Tf 11.7359 0 Td [ (\056\054) -363.508 (sparro) 25.5962 (w) 65.1068 (\054) -364.39 (recliner\051\056) -582.079 (It) -340.884 (also) -340.213 (has) -341.884 (been) -342.105 (found) -342.114 (that) ] TJ -0.00199 Tc -38.925 -11.9609 Td [ (children) -393.596 (mak) 10.187 (e) -392.687 (f) -0.59549 (e) 25.5178 (wer) -393.518 (mistak) 10.2042 (es) -392.606 (at) -392.488 (basic) -392.689 (le) 25.481 (v) 15.6052 (e) 0.18624 (l) -394.282 (o) 0.19605 (n) -392.719 (matching) ] TJ -0.001 Tc 11.952 TL T* [ (and) -269.818 (sorting) -268.899 (t) -0.28917 (asks) -268.813 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 382.2 415.748 Tm (41) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 392.163 415.748 Tm (\054) Tj ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 397.329 415.748 Tm (4) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.00199 Tc /R26 9.9626 Tf 1 0 0 1 402.306 415.748 Tm [ (\135) -268.904 (a) 0.18624 (s) -269.798 (well) -269.68 (a) 0.18624 (s) -268.896 (s) -0.58814 (imple) -269.004 (decision) -269.881 (making) ] TJ -93.4461 -11.9523 Td [ (on) -375.582 (adults\056) -686.983 (T) 0.11763 (he) -376.494 (annotation) -377.388 (o) 0.19605 (f) -376.413 (E) 0.11763 (SP) -375.604 (dataset) -377.089 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 488.391 403.796 Tm (48) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 498.345 403.796 Tm [ (\135) -375.712 (i) -0.58814 (s) -375.702 (accom\055) ] TJ -0.00199 Tc -189.486 -11.9609 Td [ (plished) -321.395 (through) -321.395 (an) -321.405 (online) -322.299 (g) 5.60691 (ame) -322.309 (which) -321.402 (demands) -323.088 (tw) 9.8807 (o) -320.496 (p) 0.19605 (lay\055) ] TJ 11.9523 TL T* [ (ers) -346.582 (i) -0.38964 (ndependently) -350.307 (propose) -348.501 (l) -0.38719 (abels) -347.481 (t) -0.38719 (o) -346.687 (one) -347.594 (image) -347.594 (for) -347.408 (match\055) ] TJ 0 Tc 11.9609 TL T* [ (ing) -370.901 (a) 0.38719 (s) -370.718 (m) 0.2083 (an) 15.8013 (y) -372.717 (w) 10.1184 (ords) -370.71 (as) -371.583 (possible) -370.879 (i) -0.19115 (n) -370.012 (a) -370.806 (certain) -370.914 (time) -370.816 (limit\056) -671.412 (It) ] TJ -0.00199 Tc 11.952 TL T* [ (is) -295.083 (demonstrated) -297.884 (that) -295.794 (humans) -297.789 (tend) -297.014 (to) -295.201 (label) -296.784 (visual) -296.696 (objects) -296.897 (a) 0.18624 (t) ] TJ -0.001 Tc T* [ (an) -278.819 (easily) -278.811 (accessible) -279.689 (s) -0.48521 (emantic) -279.694 (le) 25.6036 (v) 15.7082 (e) 0.28917 (l) -279.497 (t) -0.28917 (ermed) -279.703 (a) 0.28917 (s) -278.713 (t) -0.28917 (he) -278.821 (basic) -278.797 (le) 25.6036 (v\055) ] TJ -0.00399 Tc 11.9609 TL T* [ (el) -368.304 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 323.034 332.056 Tm (3) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.00199 Tc /R26 9.9626 Tf 1 0 0 1 328.011 332.056 Tm [ (\135\056) -662.605 (Similar) -368.201 (observ) 25.5154 (a) 0.18869 (tions) -370.007 (can) -368.306 (also) -367.39 (be) -368.296 (obtained) -369.316 (on) -368.409 (oth\055) ] TJ -0.00299 Tc -19.152 -11.9523 Td [ (er) -326.789 (human) -327.815 (annotated) -327.801 (datasets) -326.798 (s) -0.68616 (uch) -326.916 (as) -326.818 (the) -326.014 (i) -0.49011 (mage) -326.901 (captioning) ] TJ -0.00199 Tc 11.952 TL T* [ (dataset) -284.987 (MSCOCO) -285.487 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 387.572 308.152 Tm (24) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.001 Tc /R26 9.9626 Tf 1 0 0 1 397.535 308.152 Tm [ (\135) -283.193 (w) 0.09802 (hich) -285.085 (limits) -283.213 (each) -285.104 (description) -285.095 (i) -0.28917 (n) -283.301 (1) 0.29897 (6) ] TJ -88.677 -11.9609 Td [ (w) 10.0179 (o) 0.29652 (rds\056) -303.908 (This) -229.908 (limitation) -229.097 (o) 0.29652 (n) -230.016 (time) -229.084 (and) -230.883 (length) -230.888 (w) 0.09802 (ill) -228.908 (impel) -230.707 (people) ] TJ 11.9523 TL T* [ (to) -249.912 (use) -249.899 (s) -0.48521 (imple) -249.882 (b) 20.2148 (ut) -250.617 (cogniti) 25.013 (v) 15.7082 (e) -252.61 (useful) -250.581 (w) 10.0179 (o) 0.29897 (rds) -250.684 (f) -0.49501 (or) -250.703 (e) 15.6984 (xpression\056) ] TJ -0.00199 Tc 11.9523 -11.9879 Td [ (V) 59.5979 (i) -0.38964 (sion\055to\055Language) -282.487 (\050V2L\051) -279.694 (i) -0.38719 (ncludes) -281.492 (high\055le) 25.5203 (v) 15.6052 (e) 0.18624 (l) -282.301 (s) -0.58814 (emantic) ] TJ -0.001 Tc -11.9523 -11.9523 Td [ (tasks) -225.384 (w) 0.09802 (hich) -225.487 (demand) -226.386 (automatically) -227.295 (transferring) -223.698 (v) 0.29897 (isual) -226.207 (infor) 19.3889 (\055) ] TJ -0.00299 Tc 11.9609 TL T* [ (mation) -386.487 (i) -0.48766 (nto) -385.602 (human) -387.413 (language\056) -719.581 (Recently) 65.1803 (\054) -421.484 (a) -385.612 (number) -388.185 (of) -385.514 (ap\055) ] TJ 11.952 TL T* [ (proaches) -408.993 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 351.634 236.386 Tm (7) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 356.621 236.386 Tm (\054) Tj ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 363.164 236.386 Tm (14) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 373.127 236.386 Tm (\054) Tj ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 379.67 236.386 Tm (47) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.001 Tc /R26 9.9626 Tf 1 0 0 1 389.632 236.386 Tm [ (\135) -407.015 (t) -0.28917 (reat) -406.888 (V2L) -408.103 (a) 0.28917 (s) -407.006 (t) -0.28917 (ranslation) -407.099 (t) -0.28917 (asks) -406.986 (using) ] TJ -0.00199 Tc -80.7746 -11.9523 Td [ (Con) 40.0056 (v) 20.1143 (olutional) -412.392 (Neural) -410.618 (Netw) 9.91501 (orks) -409.819 (\050CNN\051) -409.78 (as) -409.78 (encoder) -411.608 (while) ] TJ 11.9609 TL T* [ (emplo) 10.1993 (y) 0.19605 (ing) -379.187 (L) 0.11763 (ong) -379.189 (Short) -378.086 (T) 69.7139 (erm) -377.589 (M) -0.11273 (emory) -378.295 (\050) -0.59304 (LSTM\051) -378.202 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R26 9.9626 Tf 1 0 0 1 508.27 212.472 Tm (9) Tj ET Q 0 0 0 1 k q 10 0 0 10 0 0 cm BT -0.00399 Tc /R26 9.9626 Tf 1 0 0 1 513.248 212.472 Tm [ (\135) -378.417 (as) -377.506 (de\055) ] TJ -0.001 Tc -204.389 -11.9523 Td [ (coder) -364.481 (to) -362.795 (directly) -362.797 (translate) -362.783 (an) -363.709 (image) -363.719 (to) -362.792 (a) -362.788 (s) -0.48521 (entence\056) -650.808 (In) -361.9 (the) ] TJ -0.00199 Tc 11.952 TL T* [ (predominant) -196.482 (neural) -195.595 (netw) 9.91501 (o) 0.19605 (rk) -194.914 (based) -194.884 (approaches\054) -208.196 (i) -0.38719 (t) -193.806 (s) -0.58814 (eems) -194.796 (t) -0.38719 (hat) ] TJ 11.9613 TL T* [ (each) -383.686 (partial) -383.517 (netw) 9.91501 (o) 0.19605 (rk) -383.708 (has) -383.588 (an) -383.706 (e) 15.5954 (xplicit) -383.5 (objecti) 24.8929 (v) 15.6052 (e) -385.514 (f) -0.51462 (unction) -383.701 (t) -0.38719 (o) ] TJ 11.952 TL T* [ (learn\054) -356.394 (ho) 25.4884 (we) 25.5154 (v) 15.6052 (e) 0.18869 (r) -338.382 (t) -0.38964 (he) -334.893 (real) -334.709 (e) 25.5154 (v) 25.4859 (a) 0.18624 (luation) -337.586 (o) 0.19605 (f) -334.812 (t) -0.38719 (he) -334.89 (generated) -336.684 (results) ] TJ T* [ (is) -276.106 (highly) -277.12 (s) -0.58569 (ubjecti) 24.8929 (v) 15.6052 (e) -278 (which) -277.998 (mak) 10.1944 (e) 0.18624 (s) -277.914 (V2L) -276.307 (challenging) -278.904 (both) -277.086 (o) 0.19605 (n) ] TJ 11.9609 TL T* [ (realization) -250.885 (and) -250.907 (e) 25.5178 (v) 25.4859 (a) 0.18869 (luation\056) ] TJ -0.001 Tc 11.9523 -11.9879 Td [ (The) -353.806 (m) 0.10783 (ain) -353.801 (p) 0.29652 (roblems) -354.612 (w) 0.09802 (hich) -353.818 (V2L) -353.916 (f) 9.50332 (aces) -353.681 (to) -352.912 (are) -352.882 (t) -0.28917 (w) 10.0179 (o) -352.897 (folds\056) ] TJ -11.9523 -11.9523 Td [ (First\054) -412.301 (t) -0.28917 (he) -380.017 (understanding) -382.684 (of) -380.799 (one) -380.897 (image) -380.892 (v) 25.5889 (a) 0.28917 (ries) -381.701 (for) -379.888 (d) 0.29897 (if) 24.8194 (ferent) ] TJ -0.00199 Tc 11.952 TL T* [ (people\056) -462.092 (The) -301.486 (d) 0.19605 (i) 24.9003 (v) 15.6052 (ersity) -300.582 (of) -301.408 (thinking) -300.602 (habit\054) -313.908 (attention) -301.518 (and) -300.582 (e) 25.5154 (v) 15.6052 (en) ] TJ 11.9613 TL T* [ (e) 15.5979 (xpression) -250.904 (leads) -249.89 (to) -249.993 (dif) 24.6993 (f) -0.59549 (erent) -249.782 (e) 15.5954 (xplanations) -252.585 (of) -249.885 (visual) -249.801 (content\056) ] TJ -0.001 Tc 11.952 TL T* [ (In) -339.318 (the) -339.294 (MSCOCO) -341.416 (dataset\054) -362.606 (each) -341.095 (image) -340.193 (is) -339.189 (associated) -341.08 (with) -339.301 (5) ] TJ 0 Tc -13.7434 -29.8887 Td (1) Tj ET Q /R34 gs Q Q q q 1 1 1 rg /a0 gs 48.406 786.422 515.188 -52.699 re f q /s5 gs /x6 Do Q q /s7 gs /x8 Do Q q /s9 gs /x10 Do Q q /s11 gs /x12 Do Q Q Q Q q 1 0 0 1 0 0 cm BT /F1 12 Tf 14.4 TL ET 1 1 1 rg n 270 47 72 14 re f* 0.5 0.5 0.5 rg BT /F2 9 Tf 10.8 TL ET BT 1 0 0 1 297 50 Tm (4962) Tj T* ET Q endstream endobj 14 0 obj << /Filter /FlateDecode /Resources << /ExtGState << /a0 << /CA 1 /ca 1 >> >> /XObject << /x18 15 0 R >> >> /Length 28 /Group << /Type /Group /S /Transparency /CS /DeviceRGB /I true >> /BBox [ 78 746 96 765 ] /Type /XObject /Subtype /Form >> stream x+O4PH/VЯ0Pp 0 endstream endobj 15 0 obj << /Filter /FlateDecode /Resources 16 0 R /Length 107 /Type /XObject /BBox [ 78 746 96 765 ] /Subtype /Form >> stream xe AC̬wʠ =p,?]%+H-
Jc "82w8VSnGW;"
endstream
endobj
16 0 obj
<<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
>>
endobj
17 0 obj
<<
/Filter /FlateDecode
/Resources <<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
/XObject <<
/x15 18 0 R
>>
>>
/Length 28
/Group <<
/Type /Group
/S /Transparency
/CS /DeviceRGB
/I true
>>
/BBox [ 67 752 84 775 ]
/Type /XObject
/Subtype /Form
>>
stream
x+O4PH/VЯ04Up
0
endstream
endobj
18 0 obj
<<
/Filter /FlateDecode
/Resources 19 0 R
/Length 228
/Type /XObject
/BBox [ 67 752 84 775 ]
/Subtype /Form
>>
stream
xeQKn!s ?FPav6R٪TS.
b];15YyR
{7QL.\:Rv/x9l+L7h%1!}i/AI(kz"U&,YO![R hg{3}4/GyYF:!w}Gn+'xJcO9i뽼_-:`
endstream
endobj
19 0 obj
<<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
>>
endobj
20 0 obj
<<
/Filter /FlateDecode
/Resources <<
/ExtGState <<
/a0 <<
/CA 1
/ca 1
>>
>>
/XObject <<
/x24 21 0 R
>>
>>
/Length 28
/Group <<
/Type /Group
/S /Transparency
/CS /DeviceRGB
/I true
>>
/BBox [ 132 751 480 772 ]
/Type /XObject
/Subtype /Form
>>
stream
x+O4PH/VЯ02Qp
0
endstream
endobj
21 0 obj
<<
/Filter /FlateDecode
/Resources 22 0 R
/Length 53223
/Type /XObject
/BBox [ 132 751 480 772 ]
/Subtype /Form
>>
stream
xtI:6%Q㨈?7rA= u%6 ?Y(WbWo{B>9
x`Znϳ|8{3?0x*z ǃ|,@:w>`c|*ϻⳅKO3`g
:_|}}><.6`Z{{3]#<_o"~:ͺgk7/Ұ@|K yp ]03ʷCmş8˽Y?>(3!Bwqs.Z8,~~=rMT̩y+/*w: uBZ_`ߵp`%M?ɝ1ɳw=vDۉy&xb4Q>d@ sg~lA