Skip to content

Latest commit

 

History

History
83 lines (57 loc) · 2.37 KB

troubleshooting.md

File metadata and controls

83 lines (57 loc) · 2.37 KB
Raw
  1. luminance problem

all the results are affected by luminance around here so i tried to remove it and make robust features by preprocessing

a. LAB negative way
removing luminance way
http://t9t9.com/60

=> tons of frame down but good performance


b. CLAHE (Contrast Limited Adaptive Histogram Equalization)
inhance the contrast by using histogram

=> good for protecting raw data but median performance

c. CLAHE (Contrast Limited Adaptive Histogram Equalization) + nagative
inhance the contrast by using histogram

=> good for protecting raw data but median performance

d. gamma correction
felt like erase important feature as well

=> median median but fast

e. gamma correction + negative
gamma between 3-4 = the best

=> if gamma getting higher, correction rate up but detection rate down

  1. ksize error

    OpenCV Error: Assertion failed (type == srcB.type() && srcA.size() == srcB.size()) #1057

    to solve this kind of error, you should make sure the sizes of images you'd like to use and filters paramaters as well

  2. on image processing

##mode, rawmode = _fromarray_typemap[typekey]

-- a/PIL/Image.py

+++ b/PIL/Image.py @@ -2207,10 +2207,14 @@ _fromarray_typemap = { # ((1, 1), "|b1"): ("1", "1"), # broken ((1, 1), "|u1"): ("L", "L"), ((1, 1), "|i1"): ("I", "I;8"),

  • ((1, 1), "<i2"): ("I", "I;16"),
  • ((1, 1), ">i2"): ("I", "I;16B"),
  • ((1, 1), "<i4"): ("I", "I;32"),
  • ((1, 1), ">i4"): ("I", "I;32B"),
  • ((1, 1), "<u2"): ("I", "I;16"),
  • ((1, 1), ">u2"): ("I", "I;16B"),
  • ((1, 1), "<i2"): ("I", "I;16S"),
  • ((1, 1), ">i2"): ("I", "I;16BS"),
  • ((1, 1), "<u4"): ("I", "I;32"),
  • ((1, 1), ">u4"): ("I", "I;32B"),
  • ((1, 1), "<i4"): ("I", "I;32S"),
  • ((1, 1), ">i4"): ("I", "I;32BS"), ((1, 1), "<f4"): ("F", "F;32F"), ((1, 1), ">f4"): ("F", "F;32BF"), ((1, 1), "<f8"): ("F", "F;64F"),

  1. for early stopping

     early = EarlyStopping(monitor='val_loss', min_delta=0, patience=5, verbose=1, mode='auto')
 callbacks = [early, lr_reducer, checkpoint]
 SupResolution.fit(x_train,
             x_train,
             validation_data=(x_test, x_test),
             epochs=30,
             batch_size=batch_size,
             callbacks=callbacks)

  2. prematured jpeg problem

https://stackoverflow.com/questions/33548956/detect-avoid-premature-end-of-jpeg-in-cv2-python