Darkflow: ๋ฐ์ดํ„ฐ ๋ณด๊ฐ•์ด ์—†์Šต๋‹ˆ๊นŒ?

์— ๋งŒ๋“  2018๋…„ 04์›” 16์ผ  ยท  4์ฝ”๋ฉ˜ํŠธ  ยท  ์ถœ์ฒ˜: thtrieu/darkflow

๋‚ด ์ƒ๊ฐ์— ์ด ์ฝ”๋“œ์—๋Š” ๋ฐ์ดํ„ฐ ํ™•์žฅ ๋ถ€๋ถ„์ด ์—†๋Š” ๋ฐ˜๋ฉด ์›๋ž˜ ๋‹คํฌ๋„ท์—๋Š” ๋…ผ๋ฌธ์— ์“ฐ์—ฌ์ง„ "๊ด‘๋ฒ”์œ„ํ•œ ๋ฐ์ดํ„ฐ ํ™•์žฅ"์ด ์žˆ์Šต๋‹ˆ๋‹ค.

์ง์ ‘ ํ•ด์•ผ ํ•˜๋‚˜์š”?

๊ฐ€์žฅ ์œ ์šฉํ•œ ๋Œ“๊ธ€

@kmsravindra , ๋‚˜๋„ ์ด ์งˆ๋ฌธ์— ๊ด€์‹ฌ์ด ์žˆ์—ˆ๊ณ  ์‹ค์ œ๋กœ ์ด ๋ฐ์ดํ„ฐ ์ฆ๊ฐ•์€ yolov2์— ์žˆ์Šต๋‹ˆ๋‹ค. ํ•จ์ˆ˜(preprocess())๋Š” yolo ํด๋”์—์„œ ๊ฐ€์ ธ์˜ค๊ณ  darkflow/darkflow/net/yolov2/data.py ๋ผ์ธ 27์˜ _batch() ํ•จ์ˆ˜์—์„œ ํ˜ธ์ถœ๋ฉ๋‹ˆ๋‹ค.

๋ชจ๋“  4 ๋Œ“๊ธ€

def preprocess(self, im, allobj = None):
    """
    Takes an image, return it as a numpy tensor that is readily
    to be fed into tfnet. If there is an accompanied annotation (allobj),
    meaning this preprocessing is serving the train process, then this
    image will be transformed with random noise to augment training data,
    using scale, translation, flipping and recolor. The accompanied
    parsed annotation (allobj) will also be modified accordingly.
    """

darkflow/net/yolo/predict.py ๋ผ์ธ 49.
๊ทธ๊ฒƒ์ด ๋ฌด์—‡์„ ์˜๋ฏธํ•œ๋‹ค๊ณ  ์ƒ๊ฐํ•ฉ๋‹ˆ๊นŒ?

@janchk ๋„ค , ๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.

@janchk , ์ด ๋ฐ์ดํ„ฐ ๋ณด๊ฐ• ์ฝ”๋“œ๊ฐ€ yolov2 ํด๋”์— ๋‚˜ํƒ€๋‚˜์ง€ ์•Š๋Š” ๊ฒƒ ๊ฐ™์Šต๋‹ˆ๊นŒ? ๋‹น์‹ ์€ ๊ทธ๊ฒƒ์— ๋Œ€ํ•ด ์•ฝ๊ฐ„์˜ ๋น›์„ ๋˜์งˆ ์ˆ˜ ์žˆ์Šต๋‹ˆ๊นŒ?

@kmsravindra , ๋‚˜๋„ ์ด ์งˆ๋ฌธ์— ๊ด€์‹ฌ์ด ์žˆ์—ˆ๊ณ  ์‹ค์ œ๋กœ ์ด ๋ฐ์ดํ„ฐ ์ฆ๊ฐ•์€ yolov2์— ์žˆ์Šต๋‹ˆ๋‹ค. ํ•จ์ˆ˜(preprocess())๋Š” yolo ํด๋”์—์„œ ๊ฐ€์ ธ์˜ค๊ณ  darkflow/darkflow/net/yolov2/data.py ๋ผ์ธ 27์˜ _batch() ํ•จ์ˆ˜์—์„œ ํ˜ธ์ถœ๋ฉ๋‹ˆ๋‹ค.

์ด ํŽ˜์ด์ง€๊ฐ€ ๋„์›€์ด ๋˜์—ˆ๋‚˜์š”?
0 / 5 - 0 ๋“ฑ๊ธ‰