profile
๐Ÿ“ฉ qtly_u@naver.com
ํƒœ๊ทธ ๋ชฉ๋ก
์ „์ฒด๋ณด๊ธฐ (53)CNN(5)Object Detection(5)YOLO(4)Stable Diffusion(3)ViTPose(3)Deep Learning(3)pose estimation(3)Vision Transformer(3)boj(3)NLP(3)VAE(3)Attention(2)DNN(2)kaggle(2)Lora(2)Computer Vision(2)git(2)Keras(2)LLM(2)๋จธ์‹ ๋Ÿฌ๋‹(2)ํฌ์Šค์ฝ” ai big data ์•„์นด๋ฐ๋ฏธ(2)Anomaly Detection(2)๋”ฅ๋Ÿฌ๋‹(2)YOLOv8(2)Resnet(2)RNN(2)Anomaly Segmentation(2)ViT(2)Bounding Box(2)LSTM(2)ํ•œ์ด์Œ(2)์บ๊ธ€(2)transformer(2)Contrastive Learning(2)ํฌ์Šค์ฝ” ai big data ์•„์นด๋ฐ๋ฏธ 20๊ธฐ(2)์ด๋ถ„ํƒ์ƒ‰(2)์นด์นด์˜ค๋ธ”๋ผ์ธ๋“œ(1)๊ฐ์ฒด ์ธ์‹(1)cold start(1)MetaFormer(1)causal inference(1)MicroNet(1)YOLO yaml(1)Soft NMS(1)Fire module(1)GPT(1)image embedding(1)ML/DL(1)inpainting(1)fewer parameters(1)๋ฐฑ์ค€ 1920๋ฒˆ(1)CEVAE(1)attention mechanism(1)๊ฒฝ๋Ÿ‰ํ™”๊ธฐ๋ฒ•(1)NMS๋ž€(1)CNN inductive bias(1)bisect(1)Object pose estimation(1)Fast(1)ํ•œ์ด์Œ๋ธ”๋ Œ๋””๋“œ๋Ÿฌ๋‹(1)latent variables(1)๋ฉ”๋‰ด ์ถ”์ฒœ(1)git ์ดˆ๊ธ‰(1)์•„์นด๋ฐ๋ฏธ 20๊ธฐ(1)Knowledge distillation(1)programmers(1)ํ•œ์ด์Œํ”„๋กœ์ ํŠธ(1)Randomforest(1)git repository(1)Seq2Seq(1)LoRA adaptation(1)Image Captioning(1)Shift-based convolution(1)parameter tuning(1)anchor box(1)bounding box to polygon(1)์ด๋ฏธ์ง€์ฒ˜๋ฆฌ(1)SOTA(1)Graph(1)ํ”„๋กœ๊ทธ๋ž˜๋จธ์Šค(1)Mixture of Experts(1)interpretability(1)sam(1)PatchCore(1)bayesian(1)๋ชจ๋ธ ๊ฒฝ๋Ÿ‰ํ™”(1)Paper(1)ํ•œ์ด์Œ ๊ณต๋ชจ์ „ ์ˆ˜์ƒ(1)ํ‚ค์ฆˆ์นดํŽ˜ ์ž…์ง€์„ ์ •(1)ORB(1)Active Shift(1)Pretrained model(1)๊ต์œก(1)Roboflow(1)Image Augmentation(1)config management(1)๊ฒฝ๋Ÿ‰ ๋„คํŠธ์›Œํฌ(1)deep learning embedding(1)UCAD(1)Latent space(1)multi-class anomaly detection(1)ํฌ์œ ๋“œ๋ฆผ(1)vision-language understanding task(1)POSTECH(1)Inception-v4(1)Custom Dataset(1)SqueezeNet(1)Non-local block(1)hybrid approaches(1)SVM(1)๋”•์…”๋„ˆ๋ฆฌ(1)ํ”ผ๋ณด๋‚˜์น˜(1)์ž„๋ฒ ๋””๋“œ ๋””๋ฐ”์ด์Šค(1)Continual Learning in Anomaly Detection(1)AE(1)์ง€์‹์ฆ๋ฅ˜๊ธฐ๋ฒ•(1)Causal Effect(1)project(1)์ผ€๋ผ์Šค(1)PyTorch(1)CNN ๊ฒฝ๋Ÿ‰ํ™”(1)AutoEncoder(1)์ปดํ“จํ„ฐ ๋น„์ „(1)knowledge decomposition(1)DP(1)๊ฒฝ๋Ÿ‰ํ™”ํˆด(1)git ๊ฐ•์˜(1)fault detection(1)Sliding Window(1)๊ตฐ์ง‘ํ™”(1)Yolo ๊ตฌ์กฐ(1)Encoder / Decoder(1)multi-view generation(1)Yolo ๋ฒ„์ „๋ณ„ ํŠน์ง•(1)ํŒŒ์ดํ† ์น˜(1)yaml(1)Collaborative Filtering(1)Hybrid recommender systems(1)AI big data ๊ต์œก(1)NeRF(1)์บ๊ธ€ ๋ถ„๋ฅ˜๋ฌธ์ œ(1)TensorFlow Lite(1)YOLO ํ•™์Šต(1)bottom up(1)German Traffic Sign Benchmark(1)GoogleNet(1)์ž…์ง€์„ ์ • ํ”„๋กœ์ ํŠธ(1)์˜์ƒ ๋ถ„๋ฅ˜(1)zeroshot prediction(1)GPT ๋‹ต๋ณ€๊ธธ์ด(1)SDS(1)ํ•œ์ด์Œ ํ›„๊ธฐ(1)python(1)๊ณผ์ ํ•ฉ ๋ฐฉ์ง€(1)Prompt Tuning(1)๋”ฅ๋Ÿฌ๋‹๋ชจ๋ธ(1)counter(1)Posco AI Big Data Academy(1)์ฝ˜ํ…์ธ  ๊ธฐ๋ฐ˜ ์ถ”์ฒœ(1)๋ฌด๋ ค20๊ธฐ(1)์ปจํ…์ธ  ๊ธฐ๋ฐ˜ ์ถ”์ฒœ(1)bottleneck(1)๊ฐ์ฒด ๊ฒ€์ถœ ๊ฒฝ๋Ÿ‰ํ™”(1)Recurrent Model(1)GPT API error(1)No-category(1)์ฝœ๋ฐฑํ•จ์ˆ˜(1)์ถ”์ฒœ์‹œ์Šคํ…œ ์‚ฌ์šฉ์˜ˆ์ œ(1)๊ฒฝ๋Ÿ‰ํ™” ๊ธฐ๋ฒ•(1)ROI(1)์ž๊ธฐ๊ณ„๋ฐœ(1)Residual block(1)offset(1)BRIEF(1)yolov5(1)์ด์ง„ํƒ์ƒ‰ ์•Œ๊ณ ๋ฆฌ์ฆ˜(1)RateLimitError(1)colab(1)์ธ๊ณต์‹ ๊ฒฝ๋ง(1)๋””๋ฐ”์ด์Šค ๊ฐ์ฒด ๊ฒ€์ถœ(1)huggingface(1)single image pose estimation(1)ํ•œ์ด์Œ ํ”„๋กœ์ ํŠธ(1)๋ถ„๋ฅ˜๊ธฐ ๋น„๊ต(1)skip connection(1)ํ•œ์ด์Œ ICT ๋ฉ˜ํ† ๋ง(1)Token Mixer(1)KL divergence derivation(1)์ž์—ฐ์–ด์ฒ˜๋ฆฌ(1)rcnn(1)ํ•˜์ดํผํŒŒ๋ผ๋ฏธํ„ฐ(1)ํฌ์Šค์ฝ” ์•„์นด๋ฐ๋ฏธ ํ›„๊ธฐ(1)์ฝ”ํ…Œ(1)Deep Neural Network(1)git ์ดˆ๋ณด(1)item-to-item(1)YOLO hyper parameter(1)ํฌ์Šค์ฝ” ์•„์นด๋ฐ๋ฏธ 20๊ธฐ(1)์˜จ๋””๋ฐ”์ด์Šค(1)Air(1)BERT(1)scalability(1)Exploitation-Exploration(1)Shift operation(1)ํ•œ์ด์Œ์œ ๋ฐ๋ฏธ(1)์ž๊ฒฉ์ฆ(1)posco(1)WGAN-GP(1)config(1)big data(1)์œ ํŠœ๋ธŒ ์ถ”์ฒœ์‹œ์Šคํ…œ(1)git ๋ช…๋ น์–ด(1)ํ…์„œํ”Œ๋กœ(1)๋จธ์‹ ๋Ÿฌ๋‹๋ถ„๋ฅ˜๋ชจ๋ธ(1)callbacks(1)Classification(1)No-Cad(1)stable diffusion webUI(1)paper-review(1)Zero123(1)wgan(1)Data Analytics(1)simon funk's SVD(1)Vanishing gradient(1)image classification(1)MVDream(1)3D Generation(1)์ถ”์ฒœ์‹œ์Šคํ…œ(1)ANN(1)Recommender System(1)feature descriptor(1)roboflow dataset(1)Industrial Image Anomaly Detection(1)pytorch JIT(1)industrial anomaly detection(1)tensorflow(1)mode collapse(1)ICT๋ฉ˜ํ† ๋ง(1)NVIDIA APEX(1)AI(1)์œ ๋ฐ๋ฏธ(1)๋จธ์‹ ๋Ÿฌ๋‹๋ถ„๋ฅ˜๊ธฐ ๋น„๊ต(1)git ์‹œ์ž‘ํ•˜๊ธฐ(1)๋”ฅ๋Ÿฌ๋‹ ๋ชจ๋ธ ๊ฒฝ๋Ÿ‰ํ™”(1)prefix tuning(1)TensorRT(1)deep learning experiments(1)Non Maximum Suppression(1)Pytorch ๊ฒฝ๋Ÿ‰ํ™”(1)MOE(1)ํฌ์Šค์ฝ” ์•„์นด๋ฐ๋ฏธ(1)1 stage detector(1)VLP(1)detection model(1)content-based recommendation(1)SyncDreamer(1)tesorflow(1)๊ตํ†ตํ‘œ์ง€ํŒ ๋ถ„๋ฅ˜(1)๋…์ผ ๊ตํ†ตํ‘œ์ง€ํŒ(1)self-attention(1)Vision-Language(1)iou(1)Linkedin ์ถ”์ฒœ์‹œ์Šคํ…œ(1)ํ•œ์ด์Œ gitlab(1)NMS(1)hydra(1)hyp.scratch-low.yaml(1)bounding box anchor box ์ฐจ์ด(1)latent-factor methods(1)augmentation parameter(1)ํฌ์Šค์ฝ” ํฌ์œ ๋“œ๋ฆผ(1)openai API(1)GPT API ์‚ฌ์šฉ(1)์ถ”์ฒœ๋ฐฉ์ •์‹(1)Negative sampling(1)OPE toxanomy(1)Yolo Architecture(1)2 stage detector(1)Bisect ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ(1)segment anything(1)์ปค์Šคํ…€ ๋ฐ์ดํ„ฐ์…‹ ํ•™์Šตํ•˜๊ธฐ(1)slow&fast(1)Natural Language Processing with Disaster Tweets(1)YOLO parameter(1)Git ๊ณต๋ถ€(1)segmentation(1)quantization(1)์ถ”์ฒœ ์•Œ๊ณ ๋ฆฌ์ฆ˜(1)youtube ์ถ”์ฒœ์‹œ์Šคํ…œ(1)inception(1)ํฌ๋กค๋ง(1)variational autoencoder(1)clip(1)GTSRB(1)github(1)Yolo version(1)๋”ฅ๋Ÿฌ๋‹๋ชจ๋ธ ๊ฒฝ๋Ÿ‰ํ™”(1)Wasserstein loss(1)roboflow object detection to segmentation(1)image-to-text generation(1)์บ๊ธ€ ๊ตํ†ตํ‘œ์ง€ํŒ๋ถ„๋ฅ˜(1)VISION(1)Binary Search(1)๋ฐฑ์ค€(1)์ด์ง„ํƒ์ƒ‰(1)MobileNetv3(1)GPI API ๋‹ต๋ณ€์ƒ์„ฑ(1)๊ณ„์‚ฐ ๊ทธ๋ž˜ํ”„(1)temporal CNN(1)on-device AI SOTA(1)selective-search(1)Embedding(1)๋”ฅ๋Ÿฌ๋‹ ํ”„๋ ˆ์ž„์›Œํฌ(1)Recommender Systems(1)stable diffusion install(1)Low-Rank Adaptation(1)2022 ํ•œ์ด์Œ ๊ณต๋ชจ์ „(1)opencv(1)150370๋ฒˆ(1)shufflenet(1)Threshold(1)ํ…์ŠคํŠธ๋ถ„์„(1)dreamfusion(1)๋ถ„๋ฅ˜๋ชจ๋ธ๋น„๊ต(1)Overlap problem(1)confidence score(1)Yolo ๋ฒ„์ „๋ณ„ ์„ฑ๋Šฅ(1)ํฌํ•ญ๊ณต๋Œ€(1)๊ณต๋ถ€(1)region-proposal(1)๋™์ ๊ณ„ํš๋ฒ•(1)detector(1)Sequence Model(1)Inductive Bias(1)2023 ๊ฐ•์„œ๊ตฌ ๋น…๋ฐ์ดํ„ฐ ํ™œ์šฉ ๊ณต๋ชจ์ „(1)dynamic programming(1)video-classification(1)DP์˜ˆ์ œ(1)Hyper-parameter(1)Long Term Dependency(1)Adapter(1)Action classification(1)Yolo series(1)Video Recognition(1)ํ˜‘์—… ํ•„ํ„ฐ๋ง(1)Yolo SOTA(1)NeuS(1)๋ชจ๋ธ ํŒŒ๋ผ๋ฏธํ„ฐ(1)gan(1)์‹œ๊ฐํ™”(1)GPT ํ† ํฐ(1)
post-thumbnail

YAML๊ณผ Hydra๋ฅผ ์ด์šฉํ•œ config ๊ด€๋ฆฌ

๋”ฅ๋Ÿฌ๋‹์ด๋‚˜ ๋จธ์‹ ๋Ÿฌ๋‹ ํ”„๋กœ์ ํŠธ๋ฅผ ํ•˜๋‹ค ๋ณด๋ฉด, ๋ชจ๋ธ์˜ config ํŒŒ์ผ์„ ํ†ตํ•ด ํ•™์Šต์— ํ•„์š”ํ•œ ๋‹ค์–‘ํ•œ ์„ค์ •์„ ์ •์˜ํ•˜๊ฒŒ ๋ฉ๋‹ˆ๋‹ค. ์ด๋Ÿฌํ•œ ์„ค์ • ํŒŒ์ผ์„ ๋งŒ๋“ค ๋•Œ ๊ฐ€์žฅ ๋งŽ์ด ์‚ฌ์šฉํ•˜๋Š” ํฌ๋งท์ด YAML์ž…๋‹ˆ๋‹ค. ๋˜ํ•œ, ์„ค์ • ํŒŒ์ผ์„ ํšจ์œจ์ ์œผ๋กœ ๊ด€๋ฆฌํ•˜๊ณ , ๋‹ค์–‘ํ•œ ์‹คํ—˜ ํ™˜๊ฒฝ์„ ์ง€์›ํ•˜๊ธฐ ์œ„ํ•ด

2024๋…„ 10์›” 2์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] NOPE: Novel Object Pose Estimation from a Single Image

NOPE: Novel Object Pose Estimation from a Single Image์€ arxiv ๊ธฐ์ค€ 23๋…„ 3์›”์— ๊ฒŒ์žฌ๋œ ํŽ˜์ดํผ์ž…๋‹ˆ๋‹ค. ํŽ˜์ดํผ ๋‚ด์šฉ์— ์•ž์„œ 6D pose estimation task๋ฅผ ์‚ดํŽด๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค. ๋จผ์ € ์ตœ๊ทผ 6D Pose esti

2024๋…„ 9์›” 13์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

Roboflow ๋ฐ์ดํ„ฐ์…‹ ์œ ํ˜•๋ณ€๊ฒฝ - Object Detection์—์„œ Instance Segmentation

๋ฐ”์šด๋”ฉ ๋ฐ•์Šค๋กœ ๋˜์–ด์žˆ๋Š” object detection ๋ฐ์ดํ„ฐ์…‹์„ ์ด์šฉํ•ด์„œ segmentation ๋ฐ์ดํ„ฐ์…‹์œผ๋กœ ๋ณ€๊ฒฝํ•˜๊ณ  ์‹ถ์„ ๋•Œ, ์ขŒ์ธก์˜ annotate๋กœ ๋“ค์–ด๊ฐ€์„œ ์ด๋ฏธ์ง€๋ฅผ ์—ฐ๋‹ค.์šฐ์ธก์˜ ํˆด๋ฐ”์—์„œ polygon tool ๋˜๋Š” smart polygon tool์„ ์‚ฌ์šฉํ•˜์—ฌ s

2024๋…„ 9์›” 5์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] PatchCore: Towards Total Recall in Industrial Anomaly Detection

Towards Total Recall in Industrial Anomaly Detection (CVPR, 2022) locally-aware patch ๋น„๊ต ๋ฐ coreset subsampling์„ ํ†ตํ•œ idustrial anomaly detection

2024๋…„ 9์›” 4์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] Unsupervised Continual Anomaly Detection with Contrastively-learned Prompt

catastrophic forgetting์—†์ด ํ•˜๋‚˜์˜ ๋ชจ๋ธ์—์„œ multi-object๋ฅผ ์ง€์†์ ์œผ๋กœ ํ•™์Šตํ•˜๊ณ , task ๊ฐ„ transfer๊ฐ€ ์ž์œ ๋กœ์šด anomaly detection ๋ชจ๋ธ\*์—ฌ๊ธฐ์„œ ๋งํ•˜๋Š” task๋Š” ๋‹ค๋ฅธ object category, anomaly det

2024๋…„ 8์›” 27์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

OpenAI API๋กœ ์—ฌ๋Ÿฌ ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ ๋‹ต๋ณ€ ๋ฝ‘์„ ๋•Œ ์ฃผ์˜* - RateLimitError, InvalidRequestError

json๋ฐ์ดํ„ฐ๋‚˜ csv์—์„œ ๊ฐ row์— ๋Œ€ํ•œ ๋‹ต๋ณ€์„ ๋ฐ›์œผ๋ ค๊ณ  ํ•  ๋•Œ, ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์—๋Ÿฌ๊ฐ€ ๋‚˜ํƒ€๋‚ฌ๋‹ค.. ํ† ํฐ ์•„๋ผ๋ ค๊ณ  ๋ฐ์ดํ„ฐ ํ•˜๋‚˜ ๋„ฃ์–ด์„œ ํ•จ์ˆ˜ ๋งŒ๋“ ๊ฑฐ ๋™์ž‘ํ•˜๋Š”์ง€ ํ™•์ธํ•œ ๋‹ค์Œ ์ „์ฒด ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•ด ๋Œ๋ ธ๋Š”๋ฐ ์—๋Ÿฌ ๋ฐœ์ƒRateLimitError: Rate limit reach

2024๋…„ 8์›” 23์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

3D Object Generation ๊ธฐ์ˆ  ๋™ํ–ฅ, ๋ชจ๋ธ ๋น„๊ต- Zero123, MVDream, SyncDreamer

์—ฐ๊ตฌ์›์—์„œ ์„ธ๋ฏธ๋‚˜ ์—ด๋ฆฌ๋Š”๊ฑฐ ๋ฉ”์ผ๋ฐ›๊ณ  ๋“ฃ๊ณ ์‹ถ์–ด์„œ ์‹ค์žฅ๋‹˜๊ป˜ ๋ง์”€๋“œ๋ฆฌ๊ณ  ๋‹ค๋ฅธ ์—ฐ๊ตฌ์‹ค ์„ธ๋ฏธ๋‚˜ ์ฐธ์„ํ•˜๊ธฐ KAIST ๋ฐ•๋ณ‘์ค€ ์—ฐ๊ตฌ์›๋‹˜์ด ์˜ค์…”์„œ 3D ์ฝ˜ํ…์ธ  ์ƒ์„ฑ ๊ธฐ์ˆ ๋™ํ–ฅ๊ณผ CVPR 2024์—์„œ ๋ฐœํ‘œํ•˜์‹  ๋…ผ๋ฌธ์„ ์†Œ๊ฐœํ•ด์ฃผ์…จ๋‹ค. ์ตœ๊ทผ multi view์˜ synthetic data ๋งŒ

2024๋…„ 8์›” 20์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

ViTPose++: Vision Transformer for Generic Body Pose Estimation

Vision Transformer๋Š” ์ปดํ“จํ„ฐ ๋น„์ „ ์ž‘์—…์—์„œ ํฐ ์ž ์žฌ๋ ฅ์„ ๋ณด์—ฌ์ฃผ์—ˆ์œผ๋ฉฐ, human body pose estimation์— ์ ์šฉ๋˜์–ด ์šฐ์ˆ˜ํ•œ ์„ฑ๋Šฅ์„ ์–ป์—ˆ์Šต๋‹ˆ๋‹ค. ๊ธฐ์กด์˜ ViTPose์—์„œ๋Š” vision transformer๋ฅผ pose estimation tas

2024๋…„ 6์›” 9์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] Inpaint Anything

Inpaint Anything ๋…ผ๋ฌธ์€ 23๋…„ 4์›”์— ๋ฐœํ‘œ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ์ด ๋…ผ๋ฌธ์€ Segment Anything Model(SAM)์„ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•œ ์ด๋ฏธ์ง€ ์ธํŽ˜์ธํŒ… ์‹œ์Šคํ…œ์„ ์†Œ๊ฐœํ•ฉ๋‹ˆ๋‹ค. ์ด ํ”„๋ ˆ์ž„์›Œํฌ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์ฃผ์š” ๊ธฐ๋Šฅ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.Remove Anything: ์‚ฌ์šฉ์ž

2024๋…„ 6์›” 7์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

GAN Mode collapse, Wasserstein Loss, Weight Clipping, Gradient Penalty

generator๊ฐ€ discriminator๊ฐ€ ๋ชป ๋งž์ถ”๋Š” ํด๋ž˜์Šค๋ฅผ ํŒŒ์•…ํ•ด์„œ ๊ทธ ํด๋ž˜์Šค๋งŒ ๊ณ„์† ์ƒ์„ฑํ•ด์„œ discriminator๊ฐ€ ์ „๋ถ€ ์˜ค๋ถ„๋ฅ˜ํ•˜๋„๋ก ํ•˜๋Š”๊ฒƒ ์ฆ‰ generator๊ฐ€ local minima์— ๊ฐ‡ํžŒ ๊ฒƒ์ด๋‹ค. Problem with BCE lossGAN์—์„œ bi

2024๋…„ 4์›” 26์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] MetaFormer Is Actually What You Need for Vision

๋ณธ ๊ธ€์—์„œ๋Š” CVPR์—์„œ 22๋…„๋„์— ๋ฐœํ‘œ๋œ MetaFormer is Actually What You Need for Vision, Yu et al.์— ๋Œ€ํ•ด ๊ฐ„๋‹จํ•˜๊ฒŒ ์ •๋ฆฌํ•˜๊ฒ ์Šต๋‹ˆ๋‹ค.๋…ผ๋ฌธ์—์„œ๋Š” ์ผ๋ฐ˜ํ™”๋œ ํŠธ๋žœ์Šคํฌ๋จธ ์•„ํ‚คํ…์ฒ˜๋ฅผ ์ œ์•ˆํ•ฉ๋‹ˆ๋‹ค.์—ฌ๊ธฐ์„œ ๊ธฐ์กด ํŠธ๋žœ์Šคํฌ๋จธ ๊ตฌ์กฐ์—์„œ Sel

2024๋…„ 3์›” 26์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] Inception v4 (2016)

Inception ์•„ํ‚คํ…์ฒ˜๋Š” ์ดˆ๊ธฐ์— GoogLeNet์œผ๋กœ ์•Œ๋ ค์ ธ ์žˆ์—ˆ์œผ๋ฉฐ, ์ดํ›„ Inception v2, Inception v3 ๋“ฑ ๋‹ค์–‘ํ•œ ๋ฒ„์ „์ด ๋ฐœํ‘œ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. Inception v4๋Š” 2016๋…„์— ์†Œ๊ฐœ๋˜์—ˆ์œผ๋ฉฐ, ๊ทธ ์ดํ›„๋กœ๋„ ๋‹ค์–‘ํ•œ ๊ฐœ์„ ์ด ์ด๋ฃจ์–ด์ง„ ๊ฒƒ์œผ๋กœ ์•Œ๋ ค์ ธ ์žˆ์Šต

2024๋…„ 3์›” 13์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

PEFT(Parameter-Efficient Fine-Tuning) ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ : ๋Œ€๊ทœ๋ชจ Pre-trained Language Model ํšจ๊ณผ์ ์œผ๋กœ ํ™œ์šฉํ•˜๊ธฐ

Pre-trained Language Model (PLM) ํšจ์œจ์ ์œผ๋กœ finetuningํ•˜๊ธฐ, PEFT ๋ฐฉ๋ฒ•๋ก  ``LoRA``, ``prompt tuning``, ``prefix tuning``

2024๋…„ 3์›” 8์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

Linux server์—์„œ Stable diffusion web-ui ์„ค์น˜ํ•˜๊ธฐ

๊นƒํ—™ ์„ค์น˜ ๋งค๋‰ด์–ผ์ฒ˜๋Ÿผ sudo ์ ‘๊ทผ์ด ๋ถˆ๊ฐ€ํ•œ server์—์„œ stable diffusion ์„ค์น˜ํ•˜๊ธฐ

2024๋…„ 3์›” 1์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation

์˜ค๋Š˜ ์†Œ๊ฐœํ•˜๋Š” BLIP(paper)๋Š”, 2022๋…„ ๋ฐœํ‘œ๋œ ๋…ผ๋ฌธ์œผ๋กœ vision-language understanding tasks์™€ generation-based tasks ๋ชจ๋‘ ์œ ์—ฐํ•˜๊ฒŒ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ๋„๋ก ์•„ํ‚คํ…์ฒ˜๋ฅผ ์„ค๊ณ„ํ•˜์˜€๊ณ , ํ•ฉ์„ฑ๋œ ์บก์…˜์„ ์ƒ์„ฑํ•˜๊ณ  ๊ธฐ์กด

2024๋…„ 1์›” 30์ผ
ยท
1๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

Stable diffusion webui ์„ค์น˜ ๋ฐ ์‹คํ–‰๋ฐฉ๋ฒ•, ์—๋Ÿฌ

github link : https://github.com/AUTOMATIC1111/stable-diffusion-webui/์œ„ ๋ ˆํฌ์ง€ํ† ๋ฆฌ๋ฅผ cloneํ•˜๊ณ  webui-user.bat ํŒŒ์ผ์„ ๋”๋ธ”ํด๋ฆญํ•˜์—ฌ ์‹คํ–‰ํ•˜๋ฉด ๋œ๋‹ค.์ด๋•Œ python์„ ์ฐพ์„ ์ˆ˜ ์—†๋‹ค๋Š” ์—๋Ÿฌ๊ฐ€

2024๋…„ 1์›” 22์ผ
ยท
9๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[paper] SlowFast Networks for Video Recognition

SlowFast Networks for Video Recognition ๋…ผ๋ฌธ ๋ฆฌ๋ทฐ

2024๋…„ 1์›” 12์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

CLIP (Contrastive Language Image Pretraining)

CLIP์€ OpenAI๊ฐ€ 2021๋…„ ๋ฐœํ‘œํ–ˆ์œผ๋ฉฐ, ์ด๋ฏธ์ง€ ์ธ์‹ ์‹œ ๋ ˆ์ด๋ธ”์ด ์•Œ๋ ค์ง€์ง€ ์•Š์€ ๋ฐ์ดํ„ฐ๋ฅผ ํšจ๊ณผ์ ์œผ๋กœ ์‚ฌ์ „ํ•™์Šต์‹œํ‚ค๋Š”๋ฐ ์‚ฌ์šฉ๋œ๋‹ค. CLIP ๋ฐฉ๋ฒ•๋ก ์˜ ํ•ต์‹ฌ์€ Image Encoder์™€ Text Encoder๋ฅผ Contrastive Learning ๋ฐฉ๋ฒ•์œผ๋กœ ํ•™์Šตํ•œ๋‹ค๋Š”

2024๋…„ 1์›” 4์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[project] ๋ฉ”๋‰ด ์ถ”์ฒœ ์‹œ์Šคํ…œ

๋‚ด๋ง˜๋Œ€๋กœ ๋งŒ๋“  ๋ฉ”๋‰ด์ถ”์ฒœ์‹œ์Šคํ…œ ์ง„ํ–‰๊ณผ์ •์„ ๊ฐ„๋žตํ•˜๊ฒŒ ์ •๋ฆฌํ•ด๋ดค๋‹ค. ํ”„๋กœ์ ํŠธ๋Š” ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘ ๋‹จ๊ณ„๋ถ€ํ„ฐ ์ถ”์ฒœ๋ฐฉ์ •์‹ ๊ตฌํ˜„, ํ‰๊ฐ€์ง€ํ‘œ ๊ณ ๋ฏผ๊นŒ์ง€ ๋‹ค์–‘ํ•œ ๊ณผ์ •์„ ๊ฑฐ์ณค๋‹ค.

2023๋…„ 12์›” 12์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท
post-thumbnail

[๊ณต๋ชจ์ „] ์„œ์šธํ˜• ํ‚ค์ฆˆ์นดํŽ˜ ์ž…์ง€์„ ์ •

๊ณต๋ชจ์ „๋ช… 2023 ๊ฐ•์„œ๊ตฌ ๋น…๋ฐ์ดํ„ฐ ํ™œ์šฉ ๊ณต๋ชจ์ „๊ณต๋ชจ๊ธฐ๊ฐ„ ~ 23. 3. 24. 18:00์ง„ํ–‰๊ธฐ๊ฐ„ 23. 3. 10. ~ 23. 3. 24. (์•ฝ 2์ฃผ)์ง„ํ–‰์ธ์› 4๋ช…๐Ÿ—“๏ธ ๋…ธ์…˜ ํŽ˜์ด์ง€๐Ÿ“„ แ„‡แ…ฎแ†ซแ„‰แ…ฅแ†จแ„‡แ…ฉแ„€แ…ฉแ„‰แ…ฅ.pdf ๐Ÿ’ฌ githubํŒ€์›๋“ค๊ณผ ํ•จ๊ป˜ ์ฃผ์ œ์™€ ๋ถ„์„ ํ”„๋กœ์„ธ์Šค๋ฅผ

2023๋…„ 12์›” 12์ผ
ยท
0๊ฐœ์˜ ๋Œ“๊ธ€
ยท