#imagenet search results

As a new CS professor in 2009, @drfeifei was inspired by math & observing how humans learn to try a new data-driven approach to training AI models. She created #ImageNet, a large Internet-scale data set. Many dismissed it as being too large and complex. But today, this practice…


📒 NOTE: IMAGENET - 14M+ images! #Note #ImageNet #AI

oderoi_'s tweet image. 📒 NOTE: 

IMAGENET - 14M+ images!

#Note #ImageNet #AI

Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients Saumyaranjan Mohanty, Konda Reddy Mopuri. Action editor: Anurag Arnab. openreview.net/forum?id=Tk78v… #imagenet #labeling #annotation


The ImageNet Challenge: Pioneering computer vision since 2012! Witness how deep learning revolutionized image recognition. #ImageNet #ComputerVision #DeepLearning #IFACET #IITKanpur

eicta_iitk's tweet image. The ImageNet Challenge: Pioneering computer vision since 2012! Witness how deep learning revolutionized image recognition. 

#ImageNet #ComputerVision #DeepLearning #IFACET #IITKanpur

Fei-Fei Li’s unveiling of #ImageNet at CVPR 2009 reshaped computer vision, unleashing millions of annotated images across thousands of categories. Yet behind this triumph lay a storm of controversy—web scraping methods and unconsented image use sparked fierce debates. Critics…


#GmP #CLIP's trippiest results of linear probe etc on #ImageNet. As they're very abstract, some deductive (-); I'll just leave the class to your imagination.😶‍🌫️🧠✨ 🤖: mandeltrippy mathematchromatic tessswirl 🤓: Yes, exactly. #AIart #AI #XAI #explainable #vision #transformer

zer0int1's tweet image. #GmP #CLIP's trippiest results of linear probe etc on #ImageNet. 

As they're very abstract, some deductive (-); I'll just leave the class to your imagination.😶‍🌫️🧠✨

🤖: mandeltrippy mathematchromatic tessswirl
🤓: Yes, exactly.

#AIart #AI #XAI #explainable #vision #transformer
zer0int1's tweet image. #GmP #CLIP's trippiest results of linear probe etc on #ImageNet. 

As they're very abstract, some deductive (-); I'll just leave the class to your imagination.😶‍🌫️🧠✨

🤖: mandeltrippy mathematchromatic tessswirl
🤓: Yes, exactly.

#AIart #AI #XAI #explainable #vision #transformer

New #J2CCertification: Chimera: State Space Models Beyond Sequences Aakash Lahoti, Tanya Marwah, Ratish Puduppully, Albert Gu openreview.net/forum?id=yv0TU… #imagenet #embeddings #attention


Let's see if we can teach #CLIP to distinguish when it is: 1. reading a word that says 'thing' -- vs. -- 2. seeing the actual 'thing'. Leveraging already existing 'close-by adversarial salience' (here: 'colors' similarity). #ImageNet #Triplet #loss #dataset #crafting #AI

zer0int1's tweet image. Let's see if we can teach #CLIP to distinguish when it is:
1. reading a word that says 'thing' 
-- vs. -- 
2. seeing the actual 'thing'. 

Leveraging already existing 'close-by adversarial salience' (here: 'colors' similarity).

#ImageNet #Triplet #loss #dataset #crafting #AI

Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients Saumyaranjan Mohanty, Konda Reddy Mopuri tmlr.infinite-conf.org/paper_pages/Tk… #imagenet #labeling #annotation

TmlrVideos's tweet image. Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients

Saumyaranjan Mohanty, Konda Reddy Mopuri

tmlr.infinite-conf.org/paper_pages/Tk…

#imagenet #labeling #annotation

Some concepts learned by #CLIP ViT-B/32 #Transcoder. Some clean -> #ImageNet class. Most broad. Presumed 'having-a-baby' feature responds to cradle, baby, pregnancy belly. Weird 'roadside danger' / unclear feature.🤔 Emergency response feature includes a lobster on its back. 😂

zer0int1's tweet image. Some concepts learned by #CLIP ViT-B/32 #Transcoder.
Some clean -> #ImageNet class.
Most broad. Presumed 'having-a-baby' feature responds to cradle, baby, pregnancy belly.
Weird 'roadside danger' / unclear feature.🤔
Emergency response feature includes a lobster on its back. 😂

To be honest - just by looking at them, I never would've guessed those to be features encoding "basketball" (#ImageNet class) in #GmP #CLIP ViT-L/14. But if you look at them with a confirmation bias lens, I guess it even makes sense.🏀 #AI #XAI #explainable #vision #transformer

zer0int1's tweet image. To be honest - just by looking at them, I never would've guessed those to be features encoding "basketball" (#ImageNet class) in #GmP #CLIP ViT-L/14.

But if you look at them with a confirmation bias lens, I guess it even makes sense.🏀

#AI #XAI #explainable #vision #transformer
zer0int1's tweet image. To be honest - just by looking at them, I never would've guessed those to be features encoding "basketball" (#ImageNet class) in #GmP #CLIP ViT-L/14.

But if you look at them with a confirmation bias lens, I guess it even makes sense.🏀

#AI #XAI #explainable #vision #transformer

Dr. #Strangeloss or: How #CLIP ViT-L/14 plummeted into a newly found local minimum of 91.4% #ImageNet / #ObjectNet accuracy. 😳😂 #loss #is #win #gradientnorm-#caterpillar #vision #transformer #finetune #AIweirdness


Quote sobre la importancia del reto ImageNet y el algoritmo AlexNet (by "Nexus: Una breve historia de las redes..." by Yuval Noah Harari) #ImageNet #AlexNet #IA #AI #reto #algoritmo #challenge emeshing.blogspot.com/2025/10/quote-…


The #ImageNet moment in #robotics ? I talked to Prof. Dr. Valada - Feels like moving from task oriented symbolic models to universal subsymbolic models... What does that mean for existing symbolic models describing the world of Robotics and Automation in general? #AI


GROOD: GRadient-Aware Out-of-Distribution Detection Mostafa ElAraby, Sabyasachi Sahoo, Yann Pequignot, Paul Novello, Liam Paull. Action editor: Yanwei Fu. openreview.net/forum?id=2V7it… #imagenet #deep #detection


Professor of Computer Science, founder of #imagenet & Stanford professor, @drfeifei insights are central to Ahura AI's goals and objectives in our technology development for learners. #ahuraai #feifeili #machinelearning #ai #artificial intelligence #stanford

AhuraAi's tweet image. Professor of Computer Science, founder of #imagenet & Stanford professor, @drfeifei insights are central to Ahura AI's goals and objectives in our technology development for learners. #ahuraai #feifeili #machinelearning #ai #artificial intelligence #stanford

MobileCLIP2: Improving Multi-Modal Reinforced Training Fartash Faghri, Pavan Kumar Anasosalu Vasu, Cem Koc et al.. Action editor: Liang-Chieh Chen. openreview.net/forum?id=WeF9z… #captioner #imagenet #captions


#ICCV2023 Dataset Quantization - Adresses the computational & memory costs of training #SoTA DNNs with large datasets - Notable Results: Using 60% of #ImageNet and 20% of #Alpaca’s instruction tuning data 👉 negligible or no performance drop! 📜 arxiv.org/abs/2308.10524 1/4

ITica007's tweet image. #ICCV2023 Dataset Quantization

- Adresses the computational & memory costs of training #SoTA DNNs with large datasets

- Notable Results: Using 60% of #ImageNet and 20% of #Alpaca’s instruction tuning data 👉 negligible or no performance drop!

📜 arxiv.org/abs/2308.10524

1/4

Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients Saumyaranjan Mohanty, Konda Reddy Mopuri. Action editor: Anurag Arnab. openreview.net/forum?id=Tk78v… #imagenet #labeling #annotation


Quote sobre la importancia del reto ImageNet y el algoritmo AlexNet (by "Nexus: Una breve historia de las redes..." by Yuval Noah Harari) #ImageNet #AlexNet #IA #AI #reto #algoritmo #challenge emeshing.blogspot.com/2025/10/quote-… via @emeshing


Quote sobre la importancia del reto ImageNet y el algoritmo AlexNet (by "Nexus: Una breve historia de las redes..." by Yuval Noah Harari) #ImageNet #AlexNet #IA #AI #reto #algoritmo #challenge emeshing.blogspot.com/2025/10/quote-…


GROOD: GRadient-Aware Out-of-Distribution Detection Mostafa ElAraby, Sabyasachi Sahoo, Yann Pequignot, Paul Novello, Liam Paull. Action editor: Yanwei Fu. openreview.net/forum?id=2V7it… #imagenet #deep #detection


7/ 2010s surge: big data + GPUs + improved algorithms. AlexNet (2012) crushed ImageNet error rates, signaling a regime change across vision, speech, and translation. 🚀 #DeepLearning #GPU #ImageNet #AI


New #J2CCertification: Chimera: State Space Models Beyond Sequences Aakash Lahoti, Tanya Marwah, Ratish Puduppully, Albert Gu openreview.net/forum?id=yv0TU… #imagenet #embeddings #attention


Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients Saumyaranjan Mohanty, Konda Reddy Mopuri tmlr.infinite-conf.org/paper_pages/Tk… #imagenet #labeling #annotation

TmlrVideos's tweet image. Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients

Saumyaranjan Mohanty, Konda Reddy Mopuri

tmlr.infinite-conf.org/paper_pages/Tk…

#imagenet #labeling #annotation

Communication-Efficient Sparse Federated Learning on Non-IID Datasets openreview.net/forum?id=kUZ6L… #sparse #imagenet #saliency


MobileCLIP2: Improving Multi-Modal Reinforced Training Fartash Faghri, Pavan Kumar Anasosalu Vasu, Cem Koc et al.. Action editor: Liang-Chieh Chen. openreview.net/forum?id=WeF9z… #captioner #imagenet #captions


Mixture of Balanced Information Bottlenecks for Long-Tailed Visual Recognition Yifan Lan, Cai xin, Jun Cheng, Shan Tan. Action editor: Yin Cui. openreview.net/forum?id=9eiAL… #bottleneck #imagenet #bottlenecks


A Curious Case of Remarkable Resilience to Gradient Attacks via Fully Convolutional and Different... Leonid Boytsov, Ameya Joshi, Filipe Condessa. Action editor: Pin-Yu Chen. openreview.net/forum?id=kt7Am… #imagenet #adversarial #adversarially


Corner Cases: How Size and Position of Objects Challenge ImageNet-Trained Models Mishal Fatima, Steffen Jung, Margret Keuper. Action editor: Sanghyuk Chun. openreview.net/forum?id=Yqf2B… #imagenet #backgrounds #background


GROOD: GRadient-Aware Out-of-Distribution Detection Mostafa ElAraby, Sabyasachi Sahoo, Yann Pequignot, Paul Novello, Liam Paull tmlr.infinite-conf.org/paper_pages/2V… #neural #imagenet #deep

TmlrVideos's tweet image. GROOD: GRadient-Aware Out-of-Distribution Detection

Mostafa ElAraby, Sabyasachi Sahoo, Yann Pequignot, Paul Novello, Liam Paull

tmlr.infinite-conf.org/paper_pages/2V…

#neural #imagenet #deep

Will the Inclusion of Generated Data Amplify Bias Across Generations in Future Image Classification Models? openreview.net/forum?id=cjZ4L… #bias #imagenet #generative


Single Train Multi Deploy on Topology Search Spaces using Kshot-Hypernet openreview.net/forum?id=WyAMI… #supernet #supernetwork #imagenet


New #FeaturedCertification: MobileCLIP2: Improving Multi-Modal Reinforced Training Fartash Faghri, Pavan Kumar Anasosalu Vasu, Cem Koc et al. openreview.net/forum?id=WeF9z… #captioner #imagenet #captions


ComFe: An Interpretable Head for Vision Transformers openreview.net/forum?id=cI4wr… #imagenet #features #vision


Coresets from Trajectories: Selecting Data via Correlation of Loss Differences openreview.net/forum?id=QY0pb… #imagenet #densenet #coresets


📒 NOTE: IMAGENET - 14M+ images! #Note #ImageNet #AI

oderoi_'s tweet image. 📒 NOTE: 

IMAGENET - 14M+ images!

#Note #ImageNet #AI

The ImageNet Challenge: Pioneering computer vision since 2012! Witness how deep learning revolutionized image recognition. #ImageNet #ComputerVision #DeepLearning #IFACET #IITKanpur

eicta_iitk's tweet image. The ImageNet Challenge: Pioneering computer vision since 2012! Witness how deep learning revolutionized image recognition. 

#ImageNet #ComputerVision #DeepLearning #IFACET #IITKanpur

DEVICE OUTLET–Refurbished Units TOPCON Maestro OCT-1 with Imagenet 6 Confiable. Compacto. Listo para usar. El primero en combinar OCT SD con fotografía de fondo en color. Estableciendo el estándar para proporcionar OCT integral completamente automatizado, #oftalmologo #imagenet

Device_Optical's tweet image. DEVICE OUTLET–Refurbished Units
TOPCON Maestro OCT-1 with Imagenet 6
Confiable. Compacto. Listo para usar.
El primero en combinar OCT SD con fotografía de fondo en color. Estableciendo el estándar para proporcionar OCT integral completamente automatizado,
#oftalmologo #imagenet

- Applicability: DQ-generated subsets can be used for training *any* neural network architecture❗ - Experiments: SoTA compression ratios without loss in model performance 🗜️ #CIFAR-10 & #ImageNet-1k 👇 3/4

ITica007's tweet image. - Applicability: DQ-generated subsets can be used for training *any* neural network architecture❗

- Experiments: SoTA compression ratios without loss in model performance 🗜️

#CIFAR-10 & #ImageNet-1k 👇

3/4

Professor of Computer Science, founder of #imagenet & Stanford professor, @drfeifei insights are central to Ahura AI's goals and objectives in our technology development for learners. #ahuraai #feifeili #machinelearning #ai #artificial intelligence #stanford

AhuraAi's tweet image. Professor of Computer Science, founder of #imagenet & Stanford professor, @drfeifei insights are central to Ahura AI's goals and objectives in our technology development for learners. #ahuraai #feifeili #machinelearning #ai #artificial intelligence #stanford

- Results: DQ achieves high compression on large datasets (#ImageNet-1k & #Alpaca) instruction tuning data, while maintaining near-original performance in both vision and language tasks 👏 - Conclusion: a universal, efficient solution for more affordable training of DNNs! 4/4

ITica007's tweet image. - Results: DQ achieves high compression on large datasets (#ImageNet-1k & #Alpaca) instruction tuning data, while maintaining near-original performance in both vision and language tasks 👏

- Conclusion: a universal, efficient solution for more affordable training of DNNs!

4/4

We found that initialising the model with #ImageNet (natural images) weights and retraining the last 40% of layers of the VGG16 backbone improved our CNN’s performance… 6/10

JessieHopson's tweet image. We found that initialising the model with #ImageNet (natural images) weights and retraining the last 40% of layers of the VGG16 backbone improved our CNN’s performance…

6/10

To be honest - just by looking at them, I never would've guessed those to be features encoding "basketball" (#ImageNet class) in #GmP #CLIP ViT-L/14. But if you look at them with a confirmation bias lens, I guess it even makes sense.🏀 #AI #XAI #explainable #vision #transformer

zer0int1's tweet image. To be honest - just by looking at them, I never would've guessed those to be features encoding "basketball" (#ImageNet class) in #GmP #CLIP ViT-L/14.

But if you look at them with a confirmation bias lens, I guess it even makes sense.🏀

#AI #XAI #explainable #vision #transformer
zer0int1's tweet image. To be honest - just by looking at them, I never would've guessed those to be features encoding "basketball" (#ImageNet class) in #GmP #CLIP ViT-L/14.

But if you look at them with a confirmation bias lens, I guess it even makes sense.🏀

#AI #XAI #explainable #vision #transformer

Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients Saumyaranjan Mohanty, Konda Reddy Mopuri tmlr.infinite-conf.org/paper_pages/Tk… #imagenet #labeling #annotation

TmlrVideos's tweet image. Coreset-Driven Re-Labeling: Tackling Noisy Annotations with Noise-Free Gradients

Saumyaranjan Mohanty, Konda Reddy Mopuri

tmlr.infinite-conf.org/paper_pages/Tk…

#imagenet #labeling #annotation

Some concepts learned by #CLIP ViT-B/32 #Transcoder. Some clean -> #ImageNet class. Most broad. Presumed 'having-a-baby' feature responds to cradle, baby, pregnancy belly. Weird 'roadside danger' / unclear feature.🤔 Emergency response feature includes a lobster on its back. 😂

zer0int1's tweet image. Some concepts learned by #CLIP ViT-B/32 #Transcoder.
Some clean -> #ImageNet class.
Most broad. Presumed 'having-a-baby' feature responds to cradle, baby, pregnancy belly.
Weird 'roadside danger' / unclear feature.🤔
Emergency response feature includes a lobster on its back. 😂

I have been meaning to do this for a long time. I have reuploaded the imagenet16 dataset, a subset of #ImageNet with 16 classes. This is a practical dataset for image classification without needing high computational resources. You can find on #Zenodo zenodo.org/record/8027520

ChristosKyrkou's tweet image. I have been meaning to do this for a long time. I have reuploaded the imagenet16 dataset, a subset of #ImageNet with 16 classes. This is a practical dataset for image classification without needing high computational resources. 

You can find on #Zenodo
zenodo.org/record/8027520

#GmP #CLIP's trippiest results of linear probe etc on #ImageNet. As they're very abstract, some deductive (-); I'll just leave the class to your imagination.😶‍🌫️🧠✨ 🤖: mandeltrippy mathematchromatic tessswirl 🤓: Yes, exactly. #AIart #AI #XAI #explainable #vision #transformer

zer0int1's tweet image. #GmP #CLIP's trippiest results of linear probe etc on #ImageNet. 

As they're very abstract, some deductive (-); I'll just leave the class to your imagination.😶‍🌫️🧠✨

🤖: mandeltrippy mathematchromatic tessswirl
🤓: Yes, exactly.

#AIart #AI #XAI #explainable #vision #transformer
zer0int1's tweet image. #GmP #CLIP's trippiest results of linear probe etc on #ImageNet. 

As they're very abstract, some deductive (-); I'll just leave the class to your imagination.😶‍🌫️🧠✨

🤖: mandeltrippy mathematchromatic tessswirl
🤓: Yes, exactly.

#AIart #AI #XAI #explainable #vision #transformer

Orthonormalising gradients improves neural network optimisation openreview.net/forum?id=Zfcmw… #imagenet #neural #gradients

TmlrSub's tweet image. Orthonormalising gradients improves neural network optimisation

openreview.net/forum?id=Zfcmw…

#imagenet #neural #gradients

Let's see if we can teach #CLIP to distinguish when it is: 1. reading a word that says 'thing' -- vs. -- 2. seeing the actual 'thing'. Leveraging already existing 'close-by adversarial salience' (here: 'colors' similarity). #ImageNet #Triplet #loss #dataset #crafting #AI

zer0int1's tweet image. Let's see if we can teach #CLIP to distinguish when it is:
1. reading a word that says 'thing' 
-- vs. -- 
2. seeing the actual 'thing'. 

Leveraging already existing 'close-by adversarial salience' (here: 'colors' similarity).

#ImageNet #Triplet #loss #dataset #crafting #AI

Training AI with ImageNet 📸🧠. ImageNet’s vast database revolutionizes visual object recognition, providing a diverse range of images for deep learning models to learn from, greatly enhancing their accuracy and understanding of the visual world. #ImageNet #MachineLearning

k_dhiwahar's tweet image. Training AI with ImageNet 📸🧠. ImageNet’s vast database revolutionizes visual object recognition, providing a diverse range of images for deep learning models to learn from, greatly enhancing their accuracy and understanding of the visual world. #ImageNet #MachineLearning

Digging deeper our study shows that distillation doesn’t affect all classes equally in a dataset. Even in mostly class-balanced datasets (e.g., #CIFAR100, #ImageNet), up to 41% of classes see significant per-class accuracy shifts post-distillation in a distilled student.

Aidamo27's tweet image. Digging deeper our study shows that distillation doesn’t affect all classes equally in a dataset. Even in mostly class-balanced datasets (e.g., #CIFAR100, #ImageNet), up to 41% of classes see significant per-class accuracy shifts post-distillation in a distilled student.

#ICCV2023 Dataset Quantization - Adresses the computational & memory costs of training #SoTA DNNs with large datasets - Notable Results: Using 60% of #ImageNet and 20% of #Alpaca’s instruction tuning data 👉 negligible or no performance drop! 📜 arxiv.org/abs/2308.10524 1/4

ITica007's tweet image. #ICCV2023 Dataset Quantization

- Adresses the computational & memory costs of training #SoTA DNNs with large datasets

- Notable Results: Using 60% of #ImageNet and 20% of #Alpaca’s instruction tuning data 👉 negligible or no performance drop!

📜 arxiv.org/abs/2308.10524

1/4

This Memorial Day, we pause to remember and honor those who gave their lives in service to our country. From all of us at ImageNet of the Treasure Coast, thank you to the heroes—and their families—who have made the ultimate sacrifice. 🇺🇸 #MemorialDay #ImageNet

ImageNetFL's tweet image. This Memorial Day, we pause to remember and honor those who gave their lives in service to our country.

From all of us at ImageNet of the Treasure Coast, thank you to the heroes—and their families—who have made the ultimate sacrifice. 🇺🇸

#MemorialDay #ImageNet

Loading...

Something went wrong.


Something went wrong.


United States Trends