You Only Look One-level Featureの解説と見せかけた物体検出のよもやま話Yusuke Uchida
?
第7回全日本コンピュータビジョン勉強会「CVPR2021読み会」(前編)の発表資料です
https://kantocv.connpass.com/event/216701/
You Only Look One-level Featureの解説と、YOLO系の雑談や、物体検出における関連する手法等を広く説明しています
1) The document proposes a training algorithm to deceive anti-spoofing verification for DNN-based speech synthesis. It trains acoustic models through an iterative process of updating the models and anti-spoofing discriminator.
2) The algorithm aims to improve speech quality by compensating for differences between natural and generated speech parameter distributions using adversarial training.
3) Evaluation results show the algorithm improves speech quality over conventional training, while also training the models to effectively deceive the anti-spoofing system. The quality gains are robust against hyperparameter settings.
You Only Look One-level Featureの解説と見せかけた物体検出のよもやま話Yusuke Uchida
?
第7回全日本コンピュータビジョン勉強会「CVPR2021読み会」(前編)の発表資料です
https://kantocv.connpass.com/event/216701/
You Only Look One-level Featureの解説と、YOLO系の雑談や、物体検出における関連する手法等を広く説明しています
1) The document proposes a training algorithm to deceive anti-spoofing verification for DNN-based speech synthesis. It trains acoustic models through an iterative process of updating the models and anti-spoofing discriminator.
2) The algorithm aims to improve speech quality by compensating for differences between natural and generated speech parameter distributions using adversarial training.
3) Evaluation results show the algorithm improves speech quality over conventional training, while also training the models to effectively deceive the anti-spoofing system. The quality gains are robust against hyperparameter settings.
The document discusses algorithms for three problems:
1. Finding the maximum value of a function of N variables in O(NlogN) time.
2. Finding the first index where a value exceeds a threshold in O(N) time.
3. Sorting N numbers in O(NlogN) time.
It also provides the time complexities of different solutions to related problems and statistics on code submissions for a programming challenge.
The document contains a binary sequence recorded over 8 days from March 24 to March 31, 2017. Each day contains a number from 1 to 8, suggesting it was part of a longer term experiment or study where binary data was collected over that period.