Dong Gong

@dgonginf

Adelaide, South Australia
donggong1.github.io
Vrijeme pridruživanja: veljača 2015.
2 fotografije ili videozapisa Fotografije i videozapisi

Tweetovi

Blokirali ste korisnika/cu @dgonginf

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @dgonginf

  1. 16. sij

    We hold a multi-level memory of the whole dataset for capturing the similarity relationship in the unsupervised domain. The multi-level memory is a hub for the complementary info from local parts, individual instances and whole domain. Check out our new work with

    Memorizing Comprehensively to Learn Adaptively: Unsupervised Cross-Domain Person Re-ID with Multi-level Memory by Xinyu Zhang et al. including
    2 korisnika označavaju da im se sviđa
    Poništi
  2. 11. sij

    When self-supervised learning meets semi-supervised learning, it does not only help feature learning but can also provide more reliable labeling signals for the unlabelled samples, via the rotation angle estimator conditioned on image class P(↖️⬆️⬇️↙️|🧐). Check out our new work:

    Semi-supervised Learning via Conditional Rotation Angle Estimation by Hai-Ming Xu et al. including
    2 proslijeđena tweeta 10 korisnika označava da im se sviđa
    Poništi
  3. 10. sij

    No real and well-aligned training data for pixel-level vision task like SR? Check our new study that generates realistic pairs in an “unsupervised” way. Gap b/w the synthesized and real data is further minimized while training for SR. Empirical studies on real data are included.

    Learning to Zoom-in via Learning to Zoom-out: Real-world Super-resolution by Generating and Adapting Degradation by et al. including
    3 korisnika označavaju da im se sviđa
    Poništi
  4. 8. sij

    Our attempt to handle the long-tail/unbalanced distributed 3D point cloud data via a memory 🧠 module memorizing the representative patterns seen during training. Both dominant and non-dominant rare cases can be handled better.

    Learning and Memorizing Representative Prototypes for 3D Point Cloud Semantic and Instance Segmentation by Tong He et al. including
    2 proslijeđena tweeta 5 korisnika označava da im se sviđa
    Poništi
  5. 16. stu 2019.

    submitted 🥳🍺

    8 korisnika označava da im se sviđa
    Poništi
  6. 17. ruj 2019.

    Vehicle re-identification by paying more attention on some specific components. A part-guided bottom-up and top-down attention module is applied.

    Part-Guided Attention Learning for Vehicle Re-Identification by Xinyu Zhang et al. including
    1 reply 3 korisnika označavaju da im se sviđa
    Poništi
  7. proslijedio/la je Tweet
    30. srp 2019.

    Regularizing Proxies with Multi-Adversarial Training for Unsupervised Domain-Adaptive Semantic Segmentation by Tong Shen et al. including

    3 proslijeđena tweeta 7 korisnika označava da im se sviđa
    Poništi
  8. 22. srp 2019.

    Our work on memory-augmented autoencoder for unsupervised anomaly detection was accepted to ! Check it out at: Code release coming soon.

    7 korisnika označava da im se sviđa
    Poništi
  9. 25. tra 2019.

    Check out our latest work for attention guided HDR imaging: .

    Read 's latest research published this week to
    1 proslijeđeni tweet 1 korisnik označava da mu se sviđa
    Poništi
  10. proslijedio/la je Tweet
    11. tra 2019.

    So apparently the cool pictures of the black hole today are from the algorithm in Bouman et al. 2016 (), a CVPR paper that has been cited a total of 11 times. Citations are not necessarily an indication of impactful work, esp. multidisciplinary work!

    14 replies 584 proslijeđena tweeta 1.922 korisnika označavaju da im se sviđa
    Poništi
  11. 5. tra 2019.

    Our new anomaly detection work with a memory-augmented autoencoder:

    Interested in ? check out the latest publication by
    13 proslijeđenih tweetova 37 korisnika označava da im se sviđa
    Poništi
  12. 28. velj 2019.

    ICML review finished. Averagely 1x☕️/paper. Feel relieved now. Need some 🍺 for lunch.

    2 korisnika označavaju da im se sviđa
    Poništi
  13. 24. velj 2019.

    4 papers accepted by . 3/4 on deep dense prediction for HDR imaging, semantic segmentation, and 3D semantic scene complication; 1/4 on variational Bayesian dropout . More pre-prints and code coming out soon. Congrats to all the great co-authors!

    1 proslijeđeni tweet 8 korisnika označava da im se sviđa
    Poništi
  14. 13. lis 2018.

    I am always interested in using AI (maybe GANs with some surrogate supervision) to generate something that can be exactly defined as something but beyond the knowledge of humans.

    ＡＩ (人工知能) が描いた裸婦画が世界的な芸術賞でグランプリを獲得 身体をもたないＡＩが描いた身体は、身体感覚が弱いひとには愉しめ、強いひとには悪酔いを覚えさせる。
    1 korisnik označava da mu se sviđa
    Poništi
  15. 7. lis 2018.

    🤣

    1:41
    Running the tests again without saving your changes
    Prikaži ovu nit
    2 korisnika označavaju da im se sviđa
    Poništi
  16. proslijedio/la je Tweet
    30. stu 2017.

    If you are interested in doing research on agents that learn by interacting with an environment, you should find this interesting. This is a pretty rich environment, covering not just vision and text, but also audio!

    Introducing HoME: a Household Multimodal Environment for AI agents. - 45,000 diverse 3D houses - Vision, Audio, Physics and Semantic (text) info - OpenAI Gym integration Paper: Repo: Site:
    Prikaži ovu nit
    151 proslijeđeni tweet 384 korisnika označavaju da im se sviđa
    Poništi
  17. proslijedio/la je Tweet
    30. srp 2017.

    ACRV team (inc. Adelaide members) just won one of the primary robotics challenges internationally:

    A great story about our team taking part in the Challenge in Japan this week!
    7 proslijeđenih tweetova 13 korisnika označava da im se sviđa
    Poništi
  18. 23. srp 2017.

    One benefit of being a sponsor: free advertisement after 10 years.

    1 korisnik označava da mu se sviđa
    Poništi

@dgonginf još nije objavio nijedan Tweet.

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·