Author Topic: Preprocess (Convert/Clean/Adjust) MOT17 Det dataset annotations to YOLOv2 format  (Read 80 times)

0 Members and 1 Guest are viewing this topic.

Offline Flavio58

Advertisement
Preprocess (Convert/Clean/Adjust) MOT17 Det dataset annotations to YOLOv2 format

MOT17 Det is a dataset for people detection challenge from MOT  (https://motchallenge.net/data/MOT17Det/). It contains 14 videos under different lighting, view, weather conditions, 7 of them are training set and another 7 are used as test set. This dataset, MOT 17Det is the improved version of MOT 16 (https://arxiv.org/pdf/1603.00831.pdf).   Dataset Statistics According to https://arxiv.org/pdf/1603.00831.pdf, MOT 16 contains ~320,000 … Continue reading Preprocess (Convert/Clean/Adjust) MOT17 Det dataset annotations to YOLOv2 format

MOT17 Det is a dataset for people detection challenge from MOT  (https://motchallenge.net/data/MOT17Det/). It contains 14 videos under different lighting, view, weather conditions, 7 of them are training set and another 7 are used as test set. This dataset, MOT 17Det is the improved version of MOT 16 (https://arxiv.org/pdf/1603.00831.pdf).


 


Dataset Statistics


According to https://arxiv.org/pdf/1603.00831.pdf, MOT 16 contains ~320,000 person annotations (Pedestrian + person_on_vehicle + static_person) Table 3.  It also contains distractor class(statues, mannikin) and reflection class(reflection of people in the mirror). These two classes could be ignored by detector, for example, if detector detects them we do not say it is false detection and if detector misses them, we do not say misdetection. That way detector can learn from only ‘clean’ annotations.


Table3.png


This dataset annotation is diferent from YOLO annotations in three ways:



  1. It contains whole video annotation in a single file

  2. It contains 12 classes

  3. its annotations are in [frm_id,seq_id,xmin,ymin,w,h,confidence,class,visibility] and not in [relative_x, relative_y, relative_w, relative_h] format.


This repo contains my script that will convert MOT17 Det annotations to YOLO format. https://github.com/Jumabek/convert_MOT16_to_yolo


I converted pedestrian, person_on_vehicle and static_person as a positive class (labeled as 0). Distraction and reflection classes are converted as don’t-know class (labeled as ‘-1’). You should customize YOLO to ignore examples with ‘-1’ class while computing the loss.


Note ‘-1’ class is neither negative nor a positive class. Hence, we should ignore those kinds of objects when computing the loss/cost-function.


Source: Preprocess (Convert/Clean/Adjust) MOT17 Det dataset annotations to YOLOv2 format


Consulente in Informatica dal 1984

Software automazione, progettazione elettronica, computer vision, intelligenza artificiale, IoT, sicurezza informatica, tecnologie di sicurezza militare, SIGINT. 

Facebook:https://www.facebook.com/flaviobernardotti58
Twitter : https://www.twitter.com/Flavio58

Cell:  +39 366 3416556

f.bernardotti@deeplearningitalia.eu

#deeplearning #computervision #embeddedboard #iot #ai

 

Related Topics

  Subject / Started by Replies Last post
0 Replies
162 Views
Last post March 17, 2018, 06:25:04 AM
by Ruggero Respigo
0 Replies
262 Views
Last post March 18, 2018, 01:23:04 AM
by Ruggero Respigo
0 Replies
100 Views
Last post June 07, 2018, 11:45:12 AM
by Ruggero Respigo
0 Replies
9 Views
Last post June 26, 2019, 10:33:51 PM
by Flavio58
0 Replies
6 Views
Last post June 30, 2019, 08:21:06 AM
by Flavio58

Sitemap 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326