跳到主要內容

OCR應用在電子元件上的辨識

 OCR Application


Example1: for SMD idenfication:

Text detect by CRAFT

 

OCR文字偵測


原始照片為網路上下載,再套上OCR文字偵測顯示結果,若有侵權請告知移除
彩色區域為偵測到文字的部份


Output 10 coordinates of corresponding text blocks

1.  144,196,286,194,287,259,145,261
2.  298,198,509,196,509,259,298,262
3.  148,262,286,262,286,321,148,321
4.  368,266,513,264,513,321,369,323
5.  145,331,472,333,471,395,145,393
6.  146,404,445,404,445,454,146,454
7.  146,453,512,453,512,502,146,502
8.  147,502,481,499,481,551,148,553
9.  148,550,614,550,614,600,148,600
10.513,600,714,600,714,648,513,648

 After image pre-processing:

 

 OCR result1:


  After image pre-processing:

 OCR result2:

 

 

Example2: for datasheet interpretation:

Text detect of TI datasheet by CRAFT


OCR results:

([[75, 11], [127, 11], [127, 31], [75, 31]], 'TEXAS', 0.999188403930061)
([[474, 4], [928, 4], [928, 32], [474, 32]], 'PACKAGE MATERIALS INFORMATION', 0.6743955072876302)
([[77, 29], [189, 29], [189, 47], [77, 47]], 'INSTRUMENTS', 0.9648384864438728)
([[37, 69], [105, 69], [105, 83], [37, 83]], 'WWM Ii.com', 0.2591723778552562)
([[851, 67], [927, 67], [927, 83], [851, 83]], '17-Dec-2020', 0.9694632061866851)
([[36, 116], [334, 116], [334, 140], [36, 140]], 'TAPE AND REEL INFORMATION', 0.9635275648694305)
([[77, 153], [209, 153], [209, 173], [77, 173]], 'REEL DIMENSIONS', 0.9998498148196943)
([[519, 153], [649, 153], [649, 171], [519, 171]], 'TAPE DIMENSIONS', 0.9997295073411925)
([[501, 177], [523, 177], [523, 193], [501, 193]], 'KO', 0.5158395828404266)
([[547, 177], [575, 177], [575, 193], [547, 193]], 'P1', 0.9455686092981291)
([[277, 265], [311, 265], [311, 281], [277, 281]], 'Reel', 0.7803041934967041)
([[263, 281], [325, 281], [325, 297], [263, 297]], 'Diameter', 0.9999848270868563)
([[482, 292], [525, 292], [525, 311], [482, 311]], 'Cavity', 0.9210923809942128)
([[575, 291], [597, 291], [597, 309], [575, 309]], 'AO', 0.6634183015102135)
([[411, 321], [777, 321], [777, 337], [411, 337]], 'Dimension designed t0 accommodate the component width', 0.867950833608363)
([[411, 337], [783, 337], [783, 355], [411, 355]], 'Dimension designed t0 accommodate the component lenglh', 0.5326388348994303)
([[410, 354], [803, 354], [803, 372], [410, 372]], 'Dimension designed to accommodate the component thickness', 0.833407196857614)
([[411, 373], [607, 373], [607, 387], [411, 387]], 'Overall width of the carrier lape', 0.8950718833069984)
([[411, 391], [661, 391], [661, 405], [411, 405]], 'Pitch between successive cavity centers', 0.8034738021947719)
([[273, 435], [379, 435], [379, 453], [273, 453]], 'Reel Width (W1)', 0.46326035723651526)
([[211, 453], [461, 453], [461, 473], [211, 473]], 'QUADRANT ASSIGNMENTS FOR PIN', 0.9963498234249092)
([[473, 457], [627, 457], [627, 473], [473, 473]], 'ORIENTATION IN TAPE', 0.9946526883413003)
([[579, 485], [677, 485], [677, 501], [579, 501]], 'Sprocket Holes', 0.8864782738301257)
([[359, 537], [381, 537], [381, 551], [359, 551]], '02', 0.79855056080002)
([[419, 573], [441, 573], [441, 589], [419, 589]], '03', 0.9665244876000211)
([[455, 573], [477, 573], [477, 587], [455, 587]], '04', 0.7811954083920432)
([[557, 571], [699, 571], [699, 587], [557, 587]], 'User Direction of Feed', 0.7642206626169022)
([[371, 635], [487, 635], [487, 651], [371, 651]], 'Pocket Quadrants', 0.9271827968630615)
([[43, 673], [199, 673], [199, 687], [43, 687]], 'All dimensions are nomina', 0.9420986204801013)
([[101, 695], [149, 695], [149, 711], [101, 711]], 'Device', 0.9991779633515514)
([[210, 693], [333, 693], [333, 713], [210, 713]], 'Package |Package', 0.7039944657559438)
([[341, 695], [371, 695], [371, 711], [341, 711]], 'Pins', 0.9980747103691101)
([[391, 695], [423, 695], [423, 711], [391, 711]], 'SPQ', 0.9748639491626118)
([[455, 695], [487, 695], [487, 711], [455, 711]], 'Reel', 0.9999948740005493)
([[519, 695], [547, 695], [547, 711], [519, 711]], 'Reel', 0.8327087175396152)
([[581, 695], [603, 695], [603, 711], [581, 711]], 'Ao', 0.43958669404583417)
([[679, 695], [701, 695], [701, 711], [679, 711]], 'Ko', 0.7125325255699436)
([[831, 695], [861, 695], [861, 711], [831, 711]], 'Pinl', 0.9302820563316345)
([[221, 709], [257, 709], [257, 729], [221, 729]], 'Type', 0.9999449253082275)
([[274, 709], [333, 709], [333, 729], [274, 729]], 'Drawing', 0.6588587928000398)
([[441, 711], [503, 711], [503, 727], [441, 727]], 'Diameter]', 0.4454272609469556)
([[515, 711], [555, 711], [555, 727], [515, 727]], 'Width', 0.876041768773338)
([[573, 711], [609, 711], [609, 727], [573, 727]], '(mm)', 0.8557351242820257)
([[623, 713], [659, 713], [659, 727], [623, 727]], '(mm)', 0.9461450576782227)
([[671, 711], [707, 711], [707, 727], [671, 727]], '(mm)', 0.9999163746833801)
([[719, 711], [757, 711], [757, 727], [719, 727]], '(mm)', 0.8036057401622272)
([[769, 711], [807, 711], [807, 727], [769, 727]], '(mm)', 0.9989970922470093)
([[815, 711], [879, 711], [879, 727], [815, 727]], 'Quadrant', 0.8622102239886623)
([[453, 727], [489, 727], [489, 741], [453, 741]], '(mm)', 0.9924454092979431)
([[503, 727], [565, 727], [565, 741], [503, 741]], 'WI (mm)', 0.7676227538381624)
([[74, 746], [172, 746], [172, 762], [74, 762]], 'CD40192BNSR', 0.9991243569694064)
([[229, 749], [251, 749], [251, 763], [229, 763]], 'SO', 0.6288756878957342)
([[293, 747], [315, 747], [315, 763], [293, 763]], 'NS', 0.9976763085484702)
([[389, 749], [425, 749], [425, 763], [389, 763]], '2000', 0.999133825302124)
([[451, 749], [489, 749], [489, 763], [451, 763]], '330.0', 0.9975234746438764)
([[519, 749], [549, 749], [549, 763], [519, 763]], '16.4', 0.9038805773204304)
([[579, 749], [603, 749], [603, 763], [579, 763]], '8.2', 0.9994752927169485)
([[625, 747], [655, 747], [655, 763], [625, 763]], '10.5', 0.958955619982524)
([[677, 749], [701, 749], [701, 763], [677, 763]], '2.5', 0.9993891985480049)
([[723, 749], [753, 749], [753, 763], [723, 763]], '12.0', 0.9999243021011353)
([[773, 749], [803, 749], [803, 763], [773, 763]], '16.0', 0.9998359680175781)
([[75, 771], [173, 771], [173, 787], [75, 787]], 'CD40193BNSR', 0.9757049564359923)
([[229, 771], [251, 771], [251, 787], [229, 787]], 'SO', 0.43406229753678177)
([[293, 771], [315, 771], [315, 787], [293, 787]], 'NS', 0.9958551861943805)
([[343, 769], [365, 769], [365, 787], [343, 787]], '16', 0.9999985670074514)
([[389, 771], [425, 771], [425, 787], [389, 787]], '2000', 0.9932358860969543)
([[451, 771], [491, 771], [491, 787], [451, 787]], '330.0', 0.9315210972130422)
([[519, 771], [549, 771], [549, 787], [519, 787]], '16.4', 0.8763715137176155)
([[579, 771], [603, 771], [603, 787], [579, 787]], '8.2', 0.9993097120080295)
([[625, 771], [655, 771], [655, 787], [625, 787]], '10.5', 0.9990535378456116)
([[677, 771], [701, 771], [701, 787], [677, 787]], '2.5', 0.9995574650566987)
([[723, 771], [755, 771], [755, 787], [723, 787]], '12.0', 0.9997783303260803)
([[771, 771], [803, 771], [803, 787], [771, 787]], '16.0', 0.9996285438537598)
([[73, 793], [175, 793], [175, 809], [73, 809]], 'CD40193BPWR', 0.7635567491218438)
([[213, 791], [265, 791], [265, 809], [213, 809]], 'TSSOP', 0.9993041963137763)
([[293, 795], [317, 795], [317, 809], [293, 809]], 'PW', 0.9953903140902491)
([[389, 793], [425, 793], [425, 809], [389, 809]], '2000', 0.9965981245040894)
([[451, 793], [491, 793], [491, 809], [451, 809]], '330.0', 0.9999706783410108)
([[519, 793], [549, 793], [549, 809], [519, 809]], '12.4', 0.7921676598867475)
([[579, 795], [603, 795], [603, 809], [579, 809]], '6.9', 0.9979926986551102)
([[627, 793], [653, 793], [653, 811], [627, 811]], '5.6', 0.9427584567925367)
([[677, 793], [701, 793], [701, 809], [677, 809]], '1.6', 0.9420010899908803)
([[727, 795], [751, 795], [751, 809], [727, 809]], '8.0', 0.9992801198328968)
([[773, 793], [803, 793], [803, 809], [773, 809]], '12.0', 0.7524788050885077)



M​ore examples:






留言

這個網誌中的熱門文章

DeepRacer

Preliminary training: deepracer-github-simapp.tar.gz Reward function: ./opt/install/sagemaker_rl_agent/lib/python3.5/site-packages/markov/environments/deepracer_env.py action = [steering_angle, throttle] TRAINING_IMAGE_SIZE = (160, 120) Plotted waypoints in vertices array of hard track Parameters: on_track, x, y, distance_from_center, car_orientation, progress, steps,                                                                          throttle, steering, track_width, waypoints, closest_waypoints Note: Above picture is from https://yanpanlau.github.io/2016/10/11/Torcs-Keras.html

增強式學習

   迴力球遊戲-ATARI     賽車遊戲DQN-ATARI 賽車遊戲-TORCS Ref:     李宏毅老師 YOUTUBE DRL 1-3 On-policy VS Off-policy On-policy     The agent learned and the agent interacting with the environment is the same     阿光自已下棋學習 Off-policy     The agent learned and the agent interacting with the environment is different     佐助下棋,阿光在旁邊看 Add a baseline:     It is possible that R is always positive     So R subtract a expectation value Policy in " Policy Gradient" means output action, like left/right/fire gamma-discounted rewards: 時間愈遠的貢獻,降低其權重 Reward Function & Action is defined in prior to training MC v.s. TD MC 蒙弟卡羅: critic after episode end : larger variance(cuz conditions differ a lot in every episode), unbiased (judge until episode end, more fair) TD: Temporal-difference approach: critic during episode :smaller variance, biased maybe atari : a3c  ...