The implementation for the paper "TTNet: Real-time temporal and spatial video analysis of table tennis" An introduction of the project could be found here (from the authors) Training much faster, achieve 120 FPS in the inference phase on a single GPU (GTX1080Ti). The implementation could achieve comparative results with the reported results in the TTNet paper. There are several limitations of the TTNet Paper (hints: Loss function, input size, and 2 more). I have implemented the task with a new approach and a new model. By default (as the above command), there are 4 modules in the TTNet model: global stage, local stage, event spotting, segmentation.

While I was scrolling on Twitter. I saw a tweet showing the word "deep learning" is plateauing on google trends. Then Yann LeCun replied that it's simply deep learning because become more normal. This reminds me of a previous blog post that I wrote. Talking about how good tech is like good design.

Last month, Sophie Wilmès, the prime minister of Belgium, appeared in an online video to tell her audience that the COVID-19 pandemic was linked to the "exploitation and destruction by humans of our natural environment." Whether or not these two existential crises are connected, the fact is that Wilmès said no such thing. Produced by an organization of climate change activists, the video was actually a deepfake, or a form of fake media created using deep learning. Deepfakes are yet another way to spread misinformation--as if there wasn't enough fake news about the pandemic already. Because new security measures consistently catch many deepfake images and videos, people may be lulled into a false sense of security and believe we have the situation under control.

Organizations today are striving to build agility and resilience to the fast-changing environment we live in. AI and machine learning innovation can help tackle these emerging challenges and enable cost efficiencies. However, organizations still encounter barriers to adopting and deploying machine learning at scale. Recently at Microsoft Ignite, Azure Machine Learning made a number of announcements that help organizations harness machine learning more easily, securely, and at scale. This includes capabilities like designer and automated machine learning UI, now generally available, that simplify machine learning for beginners and professionals alike.

Deep learning is often viewed as the exclusive domain of math PhDs and big tech companies. But as this hands-on guide demonstrates, programmers comfortable with Python can achieve impressive results in deep learning with little math background, small amounts of data, and minimal code. With fastai, the first library to provide a consistent interface to the most frequently used deep learning applications. Authors Jeremy Howard and Sylvain Gugger, the creators of fastai, show you how to train a model on a wide range of tasks using fastai and PyTorch. You'll also dive progressively further into deep learning theory to gain a complete understanding of the algorithms behind the scenes.

Few areas of industrial technology today remain untouched by artificial intelligence (AI). Fromcontrollersto ERP tofood safetyandrobots, AI is changing the technologies we use to run manufacturing and processing facilities in subtle and not-so-subtle ways. One application with a big potential to benefit from AI is quality control software. The use of smart cameras and related AI-enabled software are helping manufacturers achieve improved quality inspection at speeds, latency, and costs beyond the capabilities of human inspectors. And the timing of the arrival of these smart camera technologies is fortuitous, give the social distancing requirements of COVID-19.

The biggest AI news of 2020 so far is the success of OpenAI's monstrous new language model, GPT-3. In this post, I'm going to quickly summarize why GPT-3 has caused such a splash, before highlighting 3 consequences for individuals and companies building things with AI. Why are people excited about GPT-3? There are already lots of summary posts about GPT-3, so I won't rehash them here. For a great introduction to how the model works, check out this visual guide from the (reliably excellent) Jay Alammar.

Row A - original background, row B - background replaced with a background similar to the one in VITON dataset. We found images of a person who had a similar pose and camera perspective to the training dataset images and saw numerous artifacts present after processing (Row A). However, after removing the unusual background texture and filling the area with the same background color as in the training dataset, the received output quality was improved (although some artifacts were still present). When testing the model using more images, we discovered that the model performed semi-decently on the images similar to the ones from the training distribution and failed completely where the input was distinct enough. You can see the more successful attempts of applying the model and the typical issues we found in Fig 12.

Latent trajectory inference is a crucial problem within time-series machine learning because the identification immediately provides the interpretability of given data and the relevant systems. Some real-world data such as sequential activity of thousands of neurons [51] have higher dimension than the intrinsic dimension. Other data have lower dimension such as electrophysiological data of voltage measurements in single cells [28]. The latter problem is harder than the former because the observations may be insufficient to describe its dynamics, thus the present paper focuses on this problem to show an advantage of our method. Modeling with latent variables by neural networks have been researched after Recurrent Neural Network [53, 30] was proposed. RNN and its variants (RNNs) such as GRU [9] and LSTM [26] are the benchmark models to learn latent trajectory as the sequence of hidden units to predict or classify the observations.

Unsupervised disentangled representation learning from the unlabelled audio data, and high fidelity audio generation have become two linchpins in the machine learning research fields. However, the representation learned from an unsupervised setting does not guarantee its' usability for any downstream task at hand, which can be a wastage of the resources, if the training was conducted for that particular posterior job. Also, during the representation learning, if the model is highly biased towards the downstream task, it losses its generalisation capability which directly benefits the downstream job but the ability to scale it to other related task is lost. Therefore, to fill this gap, we propose a new autoencoder based model named "Guided Adversarial Autoencoder (GAAE)", which can learn both post-task-specific representations and the general representation capturing the factors of variation in the training data leveraging a small percentage of labelled samples; thus, makes it suitable for future related tasks. Furthermore, our proposed model can generate audio with superior quality, which is indistinguishable from the real audio samples. Hence, with the extensive experimental results, we have demonstrated that by harnessing the power of the high-fidelity audio generation, the proposed GAAE model can learn powerful representation from unlabelled dataset leveraging a fewer percentage of labelled data as supervision/guidance.