WebFace260M: A Benchmark for Million-Scale Deep Face Recognition.

In this paper, we contribute a new million-scale recognition benchmark, containing uncurated 4M identities/260M faces (WebFace260M) and cleaned 2M identities/42M faces (WebFace42M) training data, as well as an elaborately designed time-constrained evaluation protocol. Firstly, we collect 4M name lists and download 260M faces from the Internet. Then, a Cleaning Automatically utilizing Self-Training pipeline is devised to purify the tremendous WebFace260M, which is efficient and scalable. To our best knowledge, the cleaned WebFace42M is the largest public face recognition training set in the community. Referring to practical deployments, Face Recognition under Inference Time conStraint (FRUITS) protocol and a new test set with rich attributes are constructed. Moreover, we gather a large-scale masked face sub-set for biometrics assessment under COVID-19. For a comprehensive evaluation of face matchers, three recognition tasks are performed under standard, masked and unbiased settings, respectively. Equipped with this benchmark, we delve into million-scale face recognition problems. Enabled by WebFace42M, we reduce 40% failure rate on the challenging IJB-C set and rank the 3rd among 430 entries on NIST-FRVT. Even 10% data (WebFace4M) shows superior performance compared with the public training set. The proposed benchmark shows enormous potential on standard, masked and unbiased face recognition scenarios.

Prosecution records reveal pangolin trading networks in China, 2014-2019.

In a precautionary response to the current coronavirus (COVID-19) pandemic, China's Ministries permanently banned eating and trading in terrestrial wild (non-livestock) animals on 24 February 2020, and extensively updated the list of Fauna under Special State Protection (LFSSP) in 2020 and 2021, in which pangolins (Manidae spp.) were upgraded to the highest protection level. Examining 509 pangolin prosecution records from China Judgements online prior to these changes (01/01/14-31/12/19), we identified that Guangdong, Guangxi and Yunnan Provinces were hotspots for trade in whole pangolins and their scales. Interrupting trade in these three principal southern provinces would substantially fragment the pangolin trade network and reduce supply of imports from other south-east Asian countries. In the context of the revised legislation and strategies intended to prevent wildlife trade, we conclude that targeting interventions at key trade nodes could significantly reduce illegal trade in pangolins, and that this approach could also be effective with other taxa.

Analysis of CT image and clinical characteristics of 55 patients with corona virus disease 2019 and delayed diagnosis and treatment

Objective To analyze the clinical and CT imaging features of some patients diagnosed having corona virus disease 2019 (COVID-19) with delayed diagnosis and treatment in Wuhan, Hubei Province for providing a reference for diagnosis and treatment of the disease. Methods A total of 55 patients diagnosed as having COVID-19 by RT-PCT or CT admitted to Wuhan Taikang Tongji COVID-19 Specialized Hospital from 2020-02-15 to 2020-02-25 were retrospectively analyzed. Case data were collected on the first-day history, and the chest CT, blood routine and C-response protein measurements. The clinical manifestations, laboratory examinations and CT imaging features of the patients were analyzed. Results The first symptoms included fever in 33 cases (60.0%), cough in 30(54.6%) expectoration in 7(12.7%), dyspnea in 9(16.4%), chest distress in 15(27.3%), headache in 4(7.3%), fatigue in 33(60.0%), muscle soreness in 10(18.2%) and diarrhea in 7(12.7%). Most of the patients had normal white blood cell, lymphocyte and C-reactive protein levels. The imaging manifestations of CT lesions can be roughly divided into three categories: stripe shadow and consolidation as main clinical manifestations (26 cases, 47.3%), stripe shadow and consolidation mixed with ground glass opacity (10 cases, 18.2%) and patchy ground glass opacity (19 cases, 34.5%). Patients with more than two lobes involved were older, while other clinical and laboratory indexes were not closely related to imaging findings. Conclusions Most of the COVID-19 patients with delayed diagnosis and treatment may improve their clinical symptoms at the time of hospitalization with normal white blood cell, lymphocyte and C-reactive protein levels. Stripe shadow, consolidation, and patchy ground glass opacity are the major CT imaging findings. The patients with extensive lesions are older than these with localized CT imaging findings.

Chest CT of COVID-19 in patients with a negative first RT-PCR test: Comparison with patients with a positive first RT-PCR test.

To compare clinical and imaging features between patients with an initial negative reverse-transcription-polymerase chain-reaction (RT-PCR) test and patients with an initial positive RT-PCR test. CT follow-up analysis in the negative RT-PCR group is also described.Thirty-three patients with SARS-CoV-2 infection confirmed by RT-PCR, with 216 lesions upon CT, were included. Demographic information and chest CT imaging features were collected.The average age in the whole study group was 46.9 ±â€Š11.1 years, with 18 males and 15 females. Patients in the positive RT-PCR test group were more likely to have a fever than patients in the negative RT-PCR test group (85.7% vs 50%, P < .05). Lesions in the positive group were more likely to be located in the peripheral area than lesions in the negative group (83.6% vs 68.2%, P < .05). Regarding the appearance of 216 lesions, ground-glass opacities (GGOs) with consolidation (43.2%) was the most common appearance in the negative group, followed by pure GGOs (31.8%), while in the positive group, pure GGOs (32%) and GGOs with interlobular septal thickening (32.8%) were both most frequent, and the difference between them was evident (P < .05). For the follow-up analysis, the largest short-axis of a lesion was smaller upon follow-up (median size 13.6 mm vs 14 mm), albeit by a smaller margin. Pure GGOs decreased in frequency, from 31.3% to 21.3%, while consolidation increased in frequency, from 7.5% to 12.5%.The manifestations of COVID-19 in patients with a first negative RT-PCR test and patients with a positive first RT-PCR test are different to some extent. The consolidation component may increase after follow-up.