๐Ÿ“ AI & Bigdata

๐Ÿ“ AI & Bigdata/Computer Vision

[CV] Object Detection, R-CNN๊ณ„์—ด ๊ฐ„๋‹จ ์ •๋ฆฌ: R-CNN, Fast, Faster

1. R-CNN3๊ฐ€์ง€ ๋ชจ๋“ˆ (region proposal, classification, bounding box regression)์„ ๊ฐ๊ฐ ๋”ฐ๋กœ๋”ฐ๋กœ ์ˆ˜ํ–‰ํ•œ๋‹ค.region proposal ์ถ”์ถœ → 2000xCNN ์—ฐ์‚ฐ→ classification  → Bbox regressionํ•œ๊ณ„) Selective search, CNN, SVM(Multi-stage pipelines)RoI ๋งˆ๋‹ค CNN ์—ฐ์‚ฐ-> ์†๋„ ์ €ํ•˜Multi-stage pipelines-> ๋ชจ๋ธ์„ ํ•œ๋ฒˆ์— ํ•™์Šต์‹œํ‚ค์ง€ ๋ชปํ•จํ•œ๋งˆ๋””๋กœ ๋Š๋ฆฌ๊ณ  ์™ธ๋ถ€ ํˆด์— ์˜์กด์„ฑ์ด ๊ฐ•ํ•จ(Multi-stage pipelines)2. Fast R-CNNregion proposal๊ณผ feature map์„ projection →  classifi..

๐Ÿ“ AI & Bigdata/Python

[Python] Anaconda ๊ฐ€์ƒ ํ™˜๊ฒฝ ์„ค์ •

anaconda ๊ฐ€์ƒ ํ™˜๊ฒฝ ๋งŒ๋“ค๊ธฐ๊ฐ€์ƒ ํ™˜๊ฒฝ ์ƒ์„ฑํ•˜๊ธฐPython 3.7 ๋ฒ„์ „์œผ๋กœ py37๋ผ๋Š” ์ด๋ฆ„์˜ ๊ฐ€์ƒํ™˜๊ฒฝ์„ ์ƒ์„ฑ# conda create -n ๊ฐ€์ƒํ™˜๊ฒฝ์ด๋ฆ„ python=๋ฒ„์ „conda create -n py37 python=3.7  ๊ฐ€์ƒ ํ™˜๊ฒฝ ๋ฆฌ์ŠคํŠธ ํ™•์ธ#1conda env list#2conda info envs๊ฐ€์ƒ ํ™˜๊ฒฝ ํ™œ์„ฑํ™”#1. conda activate ๊ฐ€์ƒํ™˜๊ฒฝ์ด๋ฆ„conda activate py37#2.activate ๊ฐ€์ƒํ™˜๊ฒฝ์ด๋ฆ„activate py37๊ฐ€์ƒ ํ™˜๊ฒฝ ๋น„ํ™œ์„ฑํ™”#1. conda deactivate ๊ฐ€์ƒํ™˜๊ฒฝ์ด๋ฆ„conda deactivate py37#2. deactivate ๊ฐ€์ƒํ™˜๊ฒฝ์ด๋ฆ„deactivate py37๊ฐ€์ƒ ํ™˜๊ฒฝ ์‚ญ์ œconda env remove -n ๊ฐ€์ƒํ™˜๊ฒฝ์ด๋ฆ„๊ฐ€์ƒ ํ™˜๊ฒฝ์— ๋ผ์ด๋ธŒ๋Ÿฌ..

๐Ÿ“ AI & Bigdata/AI & ML & DL

[DL] Faster R-CNN

๋ณดํ˜ธ๋˜์–ด ์žˆ๋Š” ๊ธ€์ž…๋‹ˆ๋‹ค.

๐Ÿ“ AI & Bigdata/AI & ML & DL

[DL] TensorFlow- CNN ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜ ๋ชจ๋ธ(3)

๋ณดํ˜ธ๋˜์–ด ์žˆ๋Š” ๊ธ€์ž…๋‹ˆ๋‹ค.

๐Ÿ“ AI & Bigdata/AI & ML & DL

[DL] Deep Learning ํ•™์Šต ๊ธฐ๋ณธ ๊ฐœ๋…(1)

"Google ๊ณต์ธ! ํ…์„œํ”Œ๋กœ(TensorFlow) ๊ฐœ๋ฐœ์ž ์ž๊ฒฉ์ฆ ์ทจ๋“" ๋‚ด์šฉ์„ ์ธ์šฉํ–ˆ์Šต๋‹ˆ๋‹ค.Google Colab ์‚ฌ์šฉํ•˜๊ธฐ: GPU ์ง€์›(๋”ฅ๋Ÿฌ๋‹ ์‹ค์Šต ํ™˜๊ฒฝ ์ง€์›) 1. Toolํ…์„œํ”Œ๋กœ์šฐ(Tensorflow): ์‹ค๋ฌด์—์„œ ๋งŽ์ด ์‚ฌ์šฉํŒŒ์ดํ† ์น˜(pytorch): Tensorflow์˜ ํ•˜์œ„ ๋Š๋‚Œ-๊นŠ์€ ๋‚ด์šฉ, numpy์™€ ๋น„์Šท ์ผ€๋ผ์Šค(Keras): Google์ด ์ธ์ˆ˜ ํ›„ ์‚ฌ์šฉ ๊ฑฐ์˜ ์•ˆ ํ•˜๋Š” ์ถ”์„ธ์ž„ 2. ๋”ฅ๋Ÿฌ๋‹(Deep Learning) ํ•™์Šต์ง€๋„ํ•™์Šต: input, label๋น„์ง€๋„ ํ•™์Šต:inputinput: X, fature, input๋ฐ์ดํ„ฐoutput: Y. label, output๋ฐ์ดํ„ฐ(๋ณธ ๊ฐ•์˜์—์„œ๋Š” ์ง€๋„ํ•™์Šต๋งŒ ์ง„ํ–‰ํ•  ์˜ˆ์ •์ž„)  2.1 ๋”ฅ๋Ÿฌ๋‹ ํ•™์Šต(์‹คํ–‰) ์ˆœ์„œ1. import: ํ•„์š”ํ•œ ๋ชจ๋“ˆ(๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ)๋ฅผ impor..

SOIT
'๐Ÿ“ AI & Bigdata' ์นดํ…Œ๊ณ ๋ฆฌ์˜ ๊ธ€ ๋ชฉ๋ก (8 Page)