Context

Video Action Recognition dataset. Contains 5% of balanced Kinetics-400 and Kinetics-600 (Kinetics) training data as zipped folder of mp4 files.

The Kinetics dataset is a large-scale, high-quality dataset for human action recognition in videos. The dataset consists of around 500,000 video clips covering 400/600 human action classes with at least 400/600 video clips for each action class. Each video clip lasts around 10 seconds and is labeled with a single action class. The videos are collected from YouTube.

Content

More than 10000 videos in each dataset. 10-40 videos per class.

Acknowledgements

A dataset by Deepmind.

The Kinetics-600 is a large-scale action recognition dataset which consists of around 480K videos from 600 action categories. The 480K videos are divided into 390K, 30K, 60K for training, validation and test sets, respectively. Each video in the dataset is a 10-second clip of action moment annotated from raw YouTube video. It is an extensions of the Kinetics-400 dataset.

The Kinetics-400 and Kinetics-600 datasets are video understanding datasets used for learning rich and multi-scale spatiotemporal semantics from high-dimensional videos.

Dataset Preparation for IV2 Retraining

  Downloading

git clone https://github.com/qingy1337/kinetics-dataset.git cd kinetics-dataset /kinetics-dataset > git pull /kinetics-dataset > bash ./k600_downloader.sh /kinetics-dataset > bash ./k600_extractor.sh

  Reorganizing into folders

cd kinetics-dataset /kinetics-dataset > mv k600_reorganize.py ./k600/ /kinetics-dataset > cd k600 /kinetics-dataset/k600 > python k600_reorganize annotations/train.txt… See the full description on the dataset page: https://huggingface.co/datasets/cminst/Slim-Kinetics-2.

The Kinetics dataset is a large-scale, high-quality dataset for human action recognition in videos. The dataset consists of around 500,000 video clips covering 600 human action classes with at least 600 video clips for each action class. Each video clip lasts around 10 seconds and is labeled with a single action class. The videos are collected from YouTube.

Dataset

This dataset was created by Nikiforos Vagenas

Contents

Rates of hydrogen formation in the temperature range 600C to 800C are reported for three coals of widely different rank. Between 35 and 70 percent of the total hydrogen available at any one temperature disengages with first order kinetics, but the apparent activation energies calculated from the corresponding rate constants are low and vary, for the coals in question, from ca. 8 to 15 kcal/mole. Since rate control by C-H bond rupture or gaseous diffusion must be ruled out, it is concluded that the rate determining step is a function of lamellar mobility, i.e. that hydrogen forms in a bimolecular process which occurs whenever two contiguous carbon lamellae move into an appropriate configuration.

The phenyl + ethylene (C6H5 + C2H4) reaction network was explored experimentally and theoretically to understand the temperature dependence of the reaction kinetics and product distribution under various temperature and pressure conditions. The flash photolysis apparatus combining laser absorbance spectroscopy (LAS) and time-resolved molecular beam mass spectrometry (MBMS) was used to study reactions on the C8H9 potential energy surface (PES). In LAS experiments, 505.3 nm laser light selectively probed C6H5 decay, and we measured the total C6H5 consumption rate coefficients in the intermediate temperature region (400–800 K), which connects previous experiments performed in high-temperature (pyrolysis) and low-temperature (cavity-ring-down methods) regions. From the quantum chemistry calculations by Tokmakov and Lin using the G2M(RCC5)//B3LYP method, we constructed a kinetic model and estimated phenomenological pressure-dependent rate coefficients, k(T, P), with the Arkane package in the reaction mechanism generator. The MBMS experiments, performed at 600–800 K and 10–50 Torr, revealed three major product peaks: m/z = 105 (adducts, mostly 2-phenylethyl radical, but also 1-phenylethyl radical, ortho-ethyl phenyl radical, and a spiro-fused ring radical), 104 (styrene, co-product with a H atom), and 78 (benzene, co-product with C2H3 radical). Product branching ratios were predicted by the model and validated by experiments for the first time. At 600 K and 10 Torr, the yield ratio of the H-abstraction reaction (forming benzene + C2H3) is measured to be 1.1% and the H-loss channel (styrene + H) has a 2.5% yield ratio. The model predicts 1.0% for H-abstraction and 2.3% for H-loss, which is within the experimental error bars. The branching ratio and formation of styrene increase at high temperature due to the favored formally direct channel (1.0% at 600 K and 10 Torr, 5.8% at 800 K and 10 Torr in the model prediction) and the faster β-scission reactions of C8H9 isomers. The importance of pressure dependence in kinetics is verified by the increase in the yield of the stabilized adduct from radical addition from 80.2% (800 K, 10 Torr) to 88.9% (800 K, 50 Torr), at the expense of styrene + H. The pressure-dependent model developed in this work is well validated by the LAS and MBMS measurements and gives a complete picture of the C6H5 + C2H4 reaction.

Picosecond - nanosecond transient absorption (TA) spectroscopy on films as presented in the supplementary figures S3 to S7. In detail:- FigS3: TA spectra after 5 ns pump-probe delay for DR3:ICC6 with 475 nm and 680 nm excitation- FigS4: ps-ns TA spectra of DR3:IEICO-4F and DR3:ITIC after excitation at 500 nm and of DR3:PC71BM after excitation at 532 nm.- FigS5: ps-ns TA spectra of as–cast DR3:IEICO following 532 nm and 720 nm excitation.- FigS6: ps-ns TA kinetics of neat PBDB-T-2F film following excitation at 600 nm- FigS7: ps-ns TA spectra of PBDB-T-2F:IEICO with 530 nm and 800 nm excitation and of PBDB-T-2F:IEICO-4F and PBDB-T-2F:BT-CIC with 600 nm and 800 nm excitation.

Ammonia is considered as one of the promising hydrogen carriers toward a sustainable world. Plasma assisted decomposition of NH3 could provide cost- and energy-effective, low-temperature, on-demand (partial) cracking of NH3 into H2. Here, we presented a temperature-dependent plasma-chemical kinetic study to investigate the role of both electron-induced reactions and thermally induced reactions on the decomposition of NH3. We employed a plasma-chemical kinetic model (KAUSTKin), developed a plasma-chemical reaction mechanism for the numerical analysis, and introduced a temperature-controlled dielectric barrier discharge reactor for the experimental investigation using 1 mol % NH3 diluted in N2. As a result, we observed the plasma significantly lowered the cracking temperature and found that the plasma-chemical mechanism should be further improved to better predict the experiment. The commonly used rates for the key NH3 pyrolysis reaction (NH3 + M ↔ NH2 + H + M) significantly overpredicted the recombination rate at temperatures below 600 K. Furthermore, the other identified shortcomings in the available data are (i) thermal hydrazine chemistry, (ii) electron-scattering cross-section data of NxHy, (iii) electron-impact dissociation of N2, and (iv) dissociative quenching of excited states of N2. We believe that the present study will spark fundamental interest to address these shortcomings and contribute to technical advancements in plasma assisted NH3 cracking technology.