Technology

Didi Chuxinginter-city carpooling service will resume night operations on a limited basis about a month after a female passenger was allegedly murdered by an unregistered driver who accessed the platform using his fatheraccount. Called Didi Hitch, the service will return on June 15 with new safety measures, including one that only allows drivers to serve passengers of the same sex during late night hours. Didi Hitch will also began piloting a new feature later this month called &guardian mode& (not &escort mode& as reported by some publications) that automatically shares ride details with a passengeremergency contacts.

The company says Didi Hitch will resume partial nighttime service between the hours of 10PM to 12AM and 5AM to 6AM on June 15, but with what Didi says is a &tentative safety measure.& During those times, drivers will only be able to accept ride requests made by passengers of the same sex. In other words, male drivers can only accept male passengers, while female drivers can only accept female passengers.

Guardian mode will launch on June 22 as a smale-scale pilot. When a passenger turns it on, their route is automatically shared with their emergency contacts. Didi also says its platform can monitor routes in real time and &intervene in case of any unusual activity.& Another new feature, called the shared information card, will launch on the same day and display photos of both the driver and passenger and vehicle information, with the aim of allowing both parties to verify each otheridentities.

DiDi also said it will start trialing a voice recording feature for its other services, including Express, Select, ExpressPool and Minibus, in some cities.

One of the most highly-valued startups in the world, Didi now claims about 30 million daily rides and 21 million driver partners. For some passengers, however, these new safety measures may not be enough to reassure them. For one thing, last monthmurder meant that the alleged perpetrator was able to overcome several safety measures. First, he used his fatherverified driver account to access the platform. Secondly, Didi Chuxingfacial recognition technology, which it has used since 2016 to verify drivers when they first sign up and then when they log in to start shifts, failed. Didi also said that the account had received a sexual harassment complaint before the murder, though it was unclear if that was while the father or son was using it. Didi apparently failed to contact the account despite making five efforts, but the platform nevertheless continued to allow it to accept rides.

While the new safeguards might placate some users, they don&t address the core issues brought up by the murder: making sure potentially dangerous people aren&t allowed on the platform in the first place, or are dealt with promptly when complaints surface. This is not the first time a murder has been linked to a Didi driver. Two years ago a driver allegedly confessed to robbing and killing a female passenger in Shenzhen.

In a statement emailed to TechCrunch, a DiDi spokesperson said:

&After revamping our core safety functions (including enhanced facial recognition, upgraded in-app emergency buttons, and many more), we are taking cautious steps to gradually extend DiDi Hitchservice hours in response to demands from users. This recent update will increase the range of mobility options available to passengers during these hours.

As we do so, DiDi Hitch is trialing with a number of safety initiatives based on feedback and advice from riders, drivers and other members of the public in China. We understand some of the tentative initiatives that have attracted a lot of public support in this round of consultation might have never been tried before. We will closely monitor and review the results from such experiments with the public, and make continuous adjustments. Our focus is on ensuring the Hitch service is brought back in a safe and responsible way; and that users understand&and join us as we work through&the challenges involved in providing sustainable, fair and safe mobility services. The Hitch and other teams will continue to work around the clock for ever more satisfactory solutions. We will keep you posted.&

TechCrunch has asked for further information about the sex ratio of drivers on Didiplatform, since if they are predominantly male (as is the case with many taxi or ride-sharing services), then that may impact how many female passengers are able to get a late-night ride using Hitch, and will update this article if we hear back.

DiDi also confirmed today that it has placed RMB 1 million (about $150,000 ) into a fiduciary account of the Beijing Global Law Office to reward informants who are able to give information that can help police solve the case. if no information or evidence has been confirmed by police by September 1, then Didi says the money will be donated to the China Foundation for Justice and Courage, which is overseen by the Ministry of Public Security.

Farmdrop, the farmer-friendly online grocery platform based in the U.K., has picked up £10 million in new funding. New investors in this Series B round include LGT Impact Ventures (described as a growth equity investor that invests in businesses making a positive contribution to society), and Belltown Ventures, a renewable energy investment specialist with an interest in agricultural technology. Previous backer Atomico also followed on.

Founded by ex-city broker Ben Pugh in 2014, Farmdrop originally launched as a ‘click and collect& service that let you order groceries online from farmer-producers to pick up at a local collection point. However, the company has since pivoted to door-to-door delivery but with the same basic idea of a marketplace that bypasses the mass supermarkets. It claims to give consumers much fresher produce, and farmer-producers a more generous share of the retail price. Large supermarkets are known for squeezing suppliers in a bid to lower prices whilst maintaining their own profits, after all.

&The fundamental problem is that the supermarketdominance over the last fifty years has put huge amounts of downward pressure on farmgate prices,& Pugh told me when Farmdrop raised its Series A. &In this environment, the only option for producers has been to focus on yields and durability which has led to a big depreciation in the taste and nutritional quality of homegrown foods&.

To that end, Farmdrop says it now sells over 2,000 products ranging from high-welfare meat, dairy, fish, organic fruit and veg, plus household supplies and larder items. It says that 80 percent of its fresh produce is sourced directly from 208 &sustainable farmers and independent food makers& and that since 2014 the startup has generated over £5 million in revenue for small-scale British farmers.

The new capital will be used to fund further U.K. expansion after the successful launch of a second hub in Bristol and Bath in September 2017, in addition to London. &Over the next six months Farmdrop will double the total number of households it can deliver to, initially growing in the South East but with plans for a northern hub in Manchester by end of 2019,& says the company.

More broadly, Farmdrop is tapping the rise of online grocery — even if the offline to online switch is still happening quite slowly — coupled with a growing demand for high-quality produce that comes from a more ethical/sustainable supply chain (Farmdrop also uses electric vans for the last few miles of delivery). It seems to be working, too: the startup says it is now on track to achieve £10 million in annualised revenues before the end of 2018.

Adds Niklass Zennström, Skype founder and CEO of Atomico: &What we find so compelling about Farmdrop is the way they&re using technology for good. By creating a direct route to market for farmers, Farmdrop is helping to create a healthier and more efficient supply chain. We&re proud to invest in such a fantastic team and are excited about helping them scale their innovative e-grocery platform.&

Back at its re:Invent conference in November, AWS announcedits $249 DeepLens, a camera thatspecifically geared toward developers who want to build and prototype vision-centric machine learning models. The company started taking pre-orders for DeepLensa few months ago, but now the camera is actually shipping to developers.

Ahead of todaylaunch, I had a chance to attend a workshop in Seattle with DeepLenssenior product manager Jyothi Nookula and AmazonVP for AI Swami Sivasubramanian to get some hands-on time with the hardware and the software services that make it tick.

DeepLens is essentially a small Ubuntu- and Intel Atom-based computer with a built-in camera thatpowerful enough to easily run and evaluate visual machine learning models. In total, DeepLens offers about 106 GFLOPS of performance.

The hardware has all of the usual I/O ports (think Micro HDMI, USB 2.0, Audio out, etc.) to let you create prototype applications, no matter whether those are simple toy apps that send you an alert when the camera detects a bear in your backyard or an industrial application that keeps an eye on a conveyor belt in your factory. The 4 megapixel camera isn&t going to win any prizes, but itperfectly adequate for most use cases. Unsurprisingly, DeepLens is deeply integrated with the rest of AWSservices. Those include the AWS IoT service Greengrass, which you use to deploy models to DeepLens, for example, but also SageMaker, Amazonnewest tool for building machine learning models.

These integrations are also what makes getting started with the camera pretty easy. Indeed, if all you want to do is run one of the pre-built samples that AWS provides, it shouldn&t take you more than 10 minutes to set up your DeepLens and deploy one of these models to the camera. Those project templates include an object detection model that can distinguish between 20 objects (though it had some issues with toy dogs, as you can see in the image above), a style transfer example to render the camera image in the style of van Gogh, a face detection model and a model that can distinguishbetween cats and dogs and one that can recognize about 30 different actions (like playing guitar, for example). The DeepLens team is also adding a model for tracking head poses. Oh, and therealso a hot dog detection model.

But thatobviously just the beginning. As the DeepLens team stressed during our workshop, even developers who have never worked with machine learning can take the existing templates and easily extend them. In part, thatdue to the fact that a DeepLens project consists of two parts: the model and a Lambda function that runs instances of the model and lets you perform actions based on the modeloutput. And with SageMaker, AWS now offers a tool that also makes it easy to build models without having to manage the underlying infrastructure.

You could do a lot of the development on the DeepLens hardware itself, given that it is essentially a small computer, though you&re probably better off using a more powerful machine and then deploying to DeepLens using the AWS Console. If you really wanted to, you could use DeepLens as a low-powered desktop machine as it comes with Ubuntu 16.04 pre-installed.

For developers who know their way around machine learning frameworks, DeepLens makes it easy to import models from virtually all the popular tools, including Caffe, TensorFlow, MXNet and others. Itworth noting that the AWS team also built a model optimizer for MXNet models that allows them to run more efficiently on the DeepLens device.

So why did AWS buildDeepLens &The whole rationale behind DeepLens came from a simple question that we asked ourselves: How do we put machine learning in the hands of every developer,& Sivasubramanian said. &To that end, we brainstormed a number of ideas and the most promising idea was actually that developers love to build solutions as hands-on fashion on devices.& And why did AWS decide to build its own hardware instead of simply working with a partner &We had a specific customer experience in mind and wanted to make sure that the end-to-end experience is really easy,& he said. &So instead of telling somebody to go download this toolkit and then go buy this toolkit from Amazon and then wire all of these together. […] So you have to do like 20 different things, which typically takes two or three days and then you have to put the entire infrastructure together. It takes too long for somebody whoexcited about learning deep learning and building something fun.&

So if you want to get started with deep learning and build some hands-on projects, DeepLens is now available on Amazon. At $249, itnot cheap, but if you are already using AWS — and maybe even use Lambda already — itprobably the easiest way to get started with building these kind of machine learning-powered applications.