The GDELT Project

At-Scale OCR Of Television News Experiments: Results From A Sample Circa-2018 Congolese Broadcast

While our Cloud Vision montaging workflow has yielded highly robust results across the majority of broadcasts we've tested it on, it does result in some reduced accuracy in edge cases, such as the Congolese broadcast below in which it fails to discern the blurry-pixelated-black-on-green chyron text. To be fair, however, even to a human it is extremely difficult to fully discern what the text says.

FRAMEID 0 => (BONJOUR
)
-------------------------------- 
FRAMEID 1 => (DONJOUK
)
-------------------------------- 
FRAMEID 4 => (BONIO
)
-------------------------------- 
FRAMEID 5 => (BONIO
)
-------------------------------- 
FRAMEID 6 => (BONOURE
)
-------------------------------- 
FRAMEID 7 => (BONJOUR
NG
NC
)
-------------------------------- 
FRAMEID 8 => (BONJOUR
)
-------------------------------- 
FRAMEID 9 => (BONJOUR
)
-------------------------------- 
FRAMEID 10 => (DONJOUK
)
-------------------------------- 
FRAMEID 11 => (LE CONGO
)
-------------------------------- 
FRAMEID 12 => (DIRECT
)
-------------------------------- 
FRAMEID 13 => (BONIO
)
-------------------------------- 
FRAMEID 14 => (BONIO
)
-------------------------------- 
FRAMEID 15 => (BON R
)
-------------------------------- 
FRAMEID 16 => (BONI
)
-------------------------------- 
FRAMEID 17 => (BONJOUR
)
-------------------------------- 
FRAMEID 18 => (BONJOUR
)
-------------------------------- 
FRAMEID 19 => (BONJOUR
BONJOUK
)
-------------------------------- 
FRAMEID 20 => (LE CONCO
)
-------------------------------- 
FRAMEID 21 => (DIRECT
)
-------------------------------- 
FRAMEID 22 => (BONIO
)
-------------------------------- 
FRAMEID 23 => (LE CONGO
D
BONICR
)
-------------------------------- 
FRAMEID 24 => (BONJOUR
BONI R
)
-------------------------------- 
FRAMEID 25 => (BONJOUR
)
-------------------------------- 
FRAMEID 26 => (BONJOUR
)
-------------------------------- 
FRAMEID 27 => (BONJOUR
)
-------------------------------- 
FRAMEID 31 => (BONIQ
)
-------------------------------- 
FRAMEID 32 => (5
BONID
)
-------------------------------- 
FRAMEID 34 => (BONJOUR
)
-------------------------------- 
FRAMEID 35 => (BONJOUR
)
-------------------------------- 
FRAMEID 36 => (DONJOUK
)
-------------------------------- 
FRAMEID 38 => (BONJOUR
)
-------------------------------- 
FRAMEID 39 => (BONJOUR
)
-------------------------------- 
FRAMEID 40 => (BONJOUR
)
-------------------------------- 
FRAMEID 41 => (BONJOUR
)
-------------------------------- 
FRAMEID 43 => (BONJOUR
)
-------------------------------- 
FRAMEID 44 => (BONJOUR
)
-------------------------------- 
FRAMEID 45 => (DONJOUK
)
-------------------------------- 
FRAMEID 46 => (BONJOUR
)
-------------------------------- 
FRAMEID 48 => (BONJOUR
)
-------------------------------- 
FRAMEID 49 => (BONJOUR
)
-------------------------------- 
FRAMEID 50 => (BONJOUR
BONCU
)
-------------------------------- 
FRAMEID 51 => (BONJOUR
)
-------------------------------- 
FRAMEID 52 => (VIOUR
)
-------------------------------- 
FRAMEID 53 => (BONJOUR
NJOUR
)
-------------------------------- 
FRAMEID 54 => (JOOK
BONJOLIR
EOG
)
-------------------------------- 
FRAMEID 56 => (BONJOUR
DIRECT
)
-------------------------------- 
FRAMEID 57 => (BONJO
)
-------------------------------- 
FRAMEID 58 => (BONI
)
-------------------------------- 
FRAMEID 59 => (BONJ
)
-------------------------------- 
FRAMEID 60 => (BONJOUR
IF
)
-------------------------------- 
FRAMEID 61 => (BONJOUR
)
-------------------------------- 
FRAMEID 62 => (BONJOUR
)
-------------------------------- 
FRAMEID 66 => (BONICE
)
-------------------------------- 
FRAMEID 67 => (LE CONGO
BONI R
)
-------------------------------- 
FRAMEID 69 => (BONJOUR
)
-------------------------------- 
FRAMEID 70 => (BONJOUR
IF
)
-------------------------------- 
FRAMEID 71 => (DONJOUK
)
-------------------------------- 
FRAMEID 72 => (BONJOUR
)
-------------------------------- 
FRAMEID 73 => (LE CONGO
)
-------------------------------- 
FRAMEID 74 => (DIRECT
)
-------------------------------- 
FRAMEID 75 => (BONJO
)
-------------------------------- 
FRAMEID 76 => (LE CONGO
BONOUR
)
-------------------------------- 
FRAMEID 77 => (BONJOUR
)
-------------------------------- 
FRAMEID 78 => (BONJOUR
)
-------------------------------- 
FRAMEID 79 => (BONJOUR
E
)
-------------------------------- 
FRAMEID 80 => (DONJOUK
)
-------------------------------- 
FRAMEID 82 => (DIRECT
)
-------------------------------- 
FRAMEID 83 => (BONJO
)
-------------------------------- 
FRAMEID 86 => (BON'
)
-------------------------------- 
FRAMEID 87 => (BONJOUR
)
-------------------------------- 
FRAMEID 88 => (BONJOUR
)
-------------------------------- 
FRAMEID 89 => (BONJOUK
)
-------------------------------- 
FRAMEID 91 => (LE CONGO
DIRECT
09331
)
-------------------------------- 
FRAMEID 92 => (BONJO
)
-------------------------------- 
FRAMEID 93 => (BONJO
LE CONGO
)
-------------------------------- 
FRAMEID 94 => (BON R
)
-------------------------------- 
FRAMEID 95 => (NJOUR
)
-------------------------------- 
FRAMEID 96 => (BONJOUR
09:31
)
-------------------------------- 
FRAMEID 97 => (DONJOUR
09:31
)
-------------------------------- 
FRAMEID 98 => (NJOOK
)
-------------------------------- 
FRAMEID 99 => (BONJOUR
09831
)
-------------------------------- 
FRAMEID 101 => (BONJO
)
-------------------------------- 
FRAMEID 102 => (LE CONGO
BONIER
)
-------------------------------- 
FRAMEID 103 => (BONJOUR
NIOUR
)
-------------------------------- 
FRAMEID 104 => (NIOUR
)
-------------------------------- 
FRAMEID 105 => (NIOUR
)
-------------------------------- 
FRAMEID 106 => (BONJOUR
NJOUR
)
-------------------------------- 
FRAMEID 107 => (BONJOUR
)
-------------------------------- 
FRAMEID 108 => (BONJOUR
)
-------------------------------- 
FRAMEID 109 => (BONJOUR
)
-------------------------------- 
FRAMEID 110 => (BONJOUR
)
-------------------------------- 
FRAMEID 111 => (BONJOUR
BONGUE
)
-------------------------------- 
FRAMEID 112 => (UR
)
-------------------------------- 
FRAMEID 113 => (BONJOUR
NIQUR
)
-------------------------------- 
FRAMEID 114 => (BONJOUR
NJOUR
)
-------------------------------- 
FRAMEID 115 => (DONJOUK
區
)
-------------------------------- 
FRAMEID 118 => (BONJO
)
-------------------------------- 
FRAMEID 119 => (BONJO
)
-------------------------------- 
FRAMEID 120 => (5
)
-------------------------------- 
FRAMEID 121 => (5
BONJOUR
)
-------------------------------- 
FRAMEID 122 => (BONJOUR
LF
)
-------------------------------- 
FRAMEID 123 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 124 => (DONJOUK
)
-------------------------------- 
FRAMEID 127 => (BONJOL
)
-------------------------------- 
FRAMEID 128 => (BONIO
)
-------------------------------- 
FRAMEID 129 => (BONI R
)
-------------------------------- 
FRAMEID 130 => (BONJOUR
)
-------------------------------- 
FRAMEID 131 => (BONJOUR
)
-------------------------------- 
FRAMEID 132 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 133 => (DONJOOK
LE
)
-------------------------------- 
FRAMEID 135 => (DIRECT
)
-------------------------------- 
FRAMEID 136 => (BONJO
)
-------------------------------- 
FRAMEID 137 => (LE CONGO
BONIO P
)
-------------------------------- 
FRAMEID 138 => (BONJOUR
)
-------------------------------- 
FRAMEID 139 => (BONJOUR
)
-------------------------------- 
FRAMEID 140 => (BONJOUR
E
)
-------------------------------- 
FRAMEID 141 => (DONJOUK
LE
)
-------------------------------- 
FRAMEID 145 => (BONJO
)
-------------------------------- 
FRAMEID 146 => (BONICOD
)
-------------------------------- 
FRAMEID 147 => (BONJOUR
BONJOUR
)
-------------------------------- 
FRAMEID 148 => (BONJOUR
)
-------------------------------- 
FRAMEID 149 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 150 => (BONJOUK
區
)
-------------------------------- 
FRAMEID 153 => (BONJOL
)
-------------------------------- 
FRAMEID 154 => (BONJO
)
-------------------------------- 
FRAMEID 155 => (BOND
)
-------------------------------- 
FRAMEID 156 => (BONJOUR
NG
)
-------------------------------- 
FRAMEID 157 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 158 => (BONOUR
NJOUR
)
-------------------------------- 
FRAMEID 159 => (JOOK
)
-------------------------------- 
FRAMEID 160 => (BONJOUR
)
-------------------------------- 
FRAMEID 161 => (DIRECT
)
-------------------------------- 
FRAMEID 162 => (BONJOU
)
-------------------------------- 
FRAMEID 163 => (BONJO
)
-------------------------------- 
FRAMEID 164 => (BONICER
)
-------------------------------- 
FRAMEID 165 => (BONJOUR
)
-------------------------------- 
FRAMEID 166 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 167 => (BONJOUR
區
)
-------------------------------- 
FRAMEID 168 => (BONJOUK
LE
)
-------------------------------- 
FRAMEID 170 => (DIRECT
)
-------------------------------- 
FRAMEID 171 => (BONJOL
)
-------------------------------- 
FRAMEID 172 => (LE CONGO
BONICOR
)
-------------------------------- 
FRAMEID 173 => (BONJO
)
-------------------------------- 
FRAMEID 174 => (BONJOUR
)
-------------------------------- 
FRAMEID 175 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 176 => (BONJOUR
)
-------------------------------- 
FRAMEID 179 => (BONJOUR
)
-------------------------------- 
FRAMEID 180 => (BONJO!
)
-------------------------------- 
FRAMEID 181 => (BONJOUR
15
BONIC D
)
-------------------------------- 
FRAMEID 182 => (BONJOUR
BONJOUR
)
-------------------------------- 
FRAMEID 183 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 184 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 185 => (DONJOUK
區
)
-------------------------------- 
FRAMEID 188 => (BON'
)
-------------------------------- 
FRAMEID 189 => (BONJO
)
-------------------------------- 
FRAMEID 190 => (BONID
)
-------------------------------- 
FRAMEID 191 => (BONJOUR
ECONGO
)
-------------------------------- 
FRAMEID 192 => (BONJOUR
)
-------------------------------- 
FRAMEID 193 => (BONJOUR
LE
)
-------------------------------- 
FRAMEID 194 => (DONJOUK
)