Bag Week 2018: Mission Workshop’s Radian rolltop starts simple but grows piece by piece

Welcome to Bag Week 2018. Every year your faithful friends at TechCrunch spend an entire week looking at bags. Why? Because bags — often ignored but full of our important electronics — are the outward representations of our techie styles, and we put far too little thought into where we keep our most prized possessions.

I’ve always been wary of modular, rail-based bag systems. They’ve always struck me as rather military and imposing, which I suppose is kind of the point. Even Mission Workshop, whose other bags I have always enjoyed, put out one that seemed to me excessive. But they’ve tempered their style a bit and put out the Radian, a solid middle ground between their one-piece and modular systems.

The Radian is clearly aimed at the choosy, pack-loving traveler who eschews roller bags for aesthetic — which describes me to a tee. Strictly rolltop bags (originating in cyclist and outdoors circles) end up feeling restrictive in where you can stow gear, and rollers are boxy and unrefined. So the Radian takes a bit from both, with the added ability to add bits and pieces according to your needs.

What it is: Adaptable, waterproof, well-designed and not attention-grabbing

What it isn’t: Simple or lightweight

The core pack is quite streamlined, with no protruding external pockets whatsoever. There’s the main compartment — 42 liters, if you’re curious — and a cleverly hidden laptop compartment between the main one and the back pads. Both are independently lined with waterproof material (in addition to the water-resistant outer layer) and the zippers are similarly sealed. There’s also a mesh pouch hidden like the laptop area that you can pop out or stow at will.

You can roll up the rolltop and secure it with Velcro, or treat it as a big flap and snap it to a strap attached to the bottom of the bag — the straps themselves are attached with strong Velcro, so you can take them off if you’re going roll style. The “Cobra” buckle upgrade is cool but the standard plastic buckles are well made enough that you shouldn’t feel any pressure to pay the $65 to upgrade.

[gallery ids="1661276,1661280,1661279,1661282,1661283,1661278,1661277,1661272"]

Access is where things begin to diverge. Unlike most rolltop packs, you can lay the bag on the ground and unzip the top as if it were a roller, letting you access the whole space from somewhere other than the top. The flap also has its own mesh enclosure. This is extremely handy and addresses the main ergonomic issue I’ve always had with strictly top-loading bags.

In a further assimilation of rolltop qualities, there’s a secret pocket at the bottom of the bag that houses a large cloth cover that seals up the pack straps and so on, making the bag much more stowable and preventing TSA or baggage handlers from having to negotiate all that junk or bag it up themselves.

Of course, a single large compartment is rarely enough when you’re doing real traveling and need to access this document or that gadget in a hurry. So the Radian joins the Mission Workshop Arkiv modular system, which lets you add on a variety of extra pockets of various sizes and types. Just be careful that you don’t push it over the carry-on size limit (though you can always stuff the extra pockets inside temporarily).

There are six rails — two on each side and two on the back — and a handful of accessories that go on each, sliding on with sturdy metal clips. The pack I tested had two zippered side pockets, the “mini folio” and the “horizontal zip” on the back, plus a cell phone pocket for the front strap.

They’re nice but the rear ones I tried are a bit small — you’d have trouble fitting anything but a pocket paperback and a couple of energy bars in either. If I had my choice I would go with the full-size folio, one zippered and one rolltop side pocket. Then you can do away with the cell pocket, which is a bit much, and have several stowage options within reach. Plus the folio has its own rails to stick one of the small ones onto.

There’s really no need to get the separate laptop case, since the laptop compartment would honestly fit two or three. It’s a great place to store dress shirts and other items that need to stay folded up and straight.

[gallery ids="1661272,1661278,1661275,1661281,1661284,1661271"]

As far as room, the 42 liters are enough on my estimation to pack for a five-day trip — that is to say, I easily fit in five pairs of socks and underwear, five t-shirts, a sweater or two, a dress shirt, some shorts and a pair of jeans. More than that would be kind of a stretch if you were also planning on bringing things like a camera, a book or two and all the other usual travel accessories.

The main compartment has mesh areas on the side to isolate toiletries and so on, but they’re just divisions; they don’t add space. There are places for small things in the outside pockets but again, not a lot of room for much bigger than a paperback, water bottle or snack unless you spring for the folio add-on.

As for looks — the version I tested was the black camo version, obviously, which looks a little more subdued in real life than my poorly color-balanced pictures make it look. Personally I prefer the company’s flat grey over the camo and the black. Makes it even more low-profile.

In the end I think the Radian is the best option for anyone looking at Mission Workshop bags who wants a modular option, but unless you plan on swapping out pieces a lot, I’m not personally convinced that it’s better than their all-in-one bags like the Rambler and Vandal. By all means take a look at putting a Radian system together, but don’t neglect to check if any of the pre-built ones fit your needs as well.

bag week 2018

Official near-earth object plan will look into nuking asteroids and other ‘planetary defense missions’

Space is a big place, and mostly empty — but there’s no shortage of objects which, should they float our direction, could end life as we know it. A new national plan for detecting and handling such objects was proposed today, and it includes the possibility of nuclear strikes on the incoming asteroids and other “planetary defense missions.”

The plan, revealed and discussed this morning, is far from a joke — it’s just that the scales these threats operate at necessarily elevates the discourse to Hollywood levels.

It’s not so much “let’s do this” as “let’s figure out what we can do.” As such it has five major goals.

First, improve our ability to detect and track near-earth objects, or NEOs. We’ve been doing it for years, and projects like NEOWise have captured an incredible amount of these objects, ranging in size from the kind that will safely burn up in orbit, to those that might cause serious damage (like the Chelyabinsk one), to proper planet-killers.

But we often hear about NEOs being detected for the first time on near-collision courses just days before approach, or even afterwards. So the report recommends looking at how existing and new programs can be utilized to better catch these objects before they become a problem.

Second, improve our knowledge of what these objects can and have done by studying and modeling them. Not just so that we know more in general, but so that in the case of a serious incoming object we know that our predictions are sound.

Third, and this is where things go a little off the rails, we need to assess and develop NEO “deflection and disruption” technologies. After all, if a planet-killer is coming our direction, we should be able to do something, right? And perhaps it shouldn’t be the very first time we’ve tried it.

The list of proposed methods sounds like it was sourced from science fiction:

This assessment should include the most mature in-space concepts — kinetic impactors, nuclear devices, and gravity tractors for deflection, and nuclear devices for disruption — as well as less mature NEO impact prevention methods.

I wasn’t aware that space nukes and gravity tractors were our most mature concepts for this kind of thing! But again, the fact is that a city-sized object approaching at a significant fraction of the speed of light is an outlandish problem that demands outlandish solutions.

And I don’t know about you, but I’d rather we tried a space nuke once or twice on a dry run rather than do it live while Armageddon looms.

At first these assessments will be purely theoretical, of course. But in the medium and long term NASA and others are tasked with designing actual “planetary defense missions”:

This action includes preliminary designs for a gravity tractor NEO deflection mission campaign, and for a kinetic impactor mission campaign in which the spacecraft is capable of either functioning as a kinetic impactor or delivering a nuclear explosive device. For the latter case, the spacecraft would contain all systems necessary to carry and safely employ a nuclear explosive device, but would carry a mass simulator with appropriate interfaces in place of an actual nuclear device. Designs should include reconnaissance spacecraft and methods to measure the achieved deflection.

Actual flight tests “would not incorporate an actual nuclear device, or involve any nuclear explosive testing.” Not yet, anyway. It’d just be a dry run, which serves its own purposes: “Thorough flight testing of a deflection/disruption system prior to an actual planetary defense mission would substantially decrease the risk of mission failure.”

Fourth the report says that we need to collaborate on the world stage, since of course NEO strikes don’t exactly discriminate by country. So in the first place we need to strengthen our existing partnerships with countries sharing NEO-related data or studies along these lines. We should all be looking into how a potential impact could affect our country specifically, of course, since we’re the ones here — but that data should be shared and analyzed globally.

Last, “Strengthen and Routinely Exercise NEO Impact Emergency Procedures and Action Protocols.”

In other words, asteroid drills.

But it isn’t just stuff like “here’s where Boulder residents should evacuate to in case of impact.” As the document points out, NEO impacts are a unique sort of emergency event.

Response and mitigation actions cannot be made routine to the same degree that they are for other natural disasters such as hurricanes. Rather, establishing and exercising thresholds and protocols will aid agencies in preparing options and recommending courses of action.

The report recommends exploring some realistic scenarios based on objects or situations we know to exist and seeing how they might play out — who will need to get involved? How will data be shared? Who is in charge of coordinating the agencies if it’s a domestic impact versus a foreign one? (See Shin Godzilla for a surprisingly good example of bureaucratic paralysis in the face of an unknown threat.)

It’s strange to think that we’re really contemplating these issues, but it’s a lot better than sitting on our hands waiting for the Big One to hit. You can read the rest of the recommendations here.

Bag Week 2018: Waxed canvas bags from Filson, Ona, Croots and more

If you’re looking for a good jacket or bag, you have your choice of materials: leather, heavy nylon, waterproof synthetic weaves like Gore-Tex… but for my money (and not a little of it either) the king of them all is waxed canvas. Pliant yet protective, wind and water–resistant but breathable, handsome to start but grows a character of its own, waxed canvas strikes, for me, the perfect balance of attributes. I drape myself in it, and in the case of bags, drape it from myself.

The main caveat is that it is not is cheap — sure, you can get a bag for $30 or $40 on Amazon, but if you want something that will live for years and years and get better with age, you’re going to be spending quite a bit more than that.

The bags here are expensive, but like leather the craftsmanship and material quality matter a great deal in whether you end up with an item that deteriorates steadily or comes into its own. Like so many things, you get what you pay for — up to a certain point, of course.

I’ve collected bags from a variety of producers and tried them all for the last few months during everyday use and trips out of town. I focused on the “fits a medium-size laptop with room for a couple books and a camera” size, but many of these makers have plenty of variety to choose from.

Check the galleries under each bag to see examples of anything I pick out as nice or irritating. (The galleries are all really tall because of a bug in our system. Don’t worry about it.)

ONA Union Street ($299) and Brixton ($289)

Pros: Rigidity and padding, customizable dividers, nice snaps

Cons: Cheap-feeling interior, bulky, could be waxier

Ona’s bags, at least these, are aimed more at the laptop-camera combo than others, with extra padding and internal dividers for bodies and extra lenses.

I reviewed the Union Street years and years ago during a previous bag week and liked it so much that I decided to buy one. It’s the larger of these two bags, fitting a 15-inch laptop and a DSLR with an extra lens or two small ones.

Not only is the whole interior lined with padding, but the dividers are padded and the main flap itself has a sturdiness that has helped protect my gear against drops and kicks. The bottom, although it is also padded and feels soft, has lived through years of scooting around and placement on rough terrain.

I like the spring-powered self-locking snaps, though when I first got the bag I was convinced they’d be the first thing to fail. Seven years and thousands of snaps later, they’re still going strong, and when I was worried one was failing (it didn’t), Ona gladly sent me a replacement.

It was my standby for a long time, and I still have it. It has aged well in some ways, not so well in others — its waxed front has survived years of scratches and slides along the floor and is marvelously smooth and still water resistant. I don’t know how they did it. On the other hand, some areas have worn holes and the magnet that holds the back flap shut (a smart idea) eventually burrowed its way out.

The newer one feels very lightly waxed, but I know it’s in there. That said, if you want the full waxy look and feel, it could use a bit more. It’s really a matter of taste.

[gallery ids="1660239,1660237,1660238,1660236,1660241,1660240"]

The inside is the weakest link. The fuzzy plush interior feels cheap to me (though it’s undeniably protective), there are no internal pockets, and repeated sticking and unsticking of the Velcro dividers wears the material down in places. Although being able to customize the interior space is invaluable for photographers specifically, a couple strong decisions inside would make it a better all-purpose bag, in my opinion.

The Brixton is the Union Street’s smaller sibling, fitting a 13-inch laptop and a bit less camera-wise. They share many qualities, including price (only a $10 difference) and ultimately the decision is one of what you need rather than which is better.

For me it’s a toss-up. I like the open, separate pockets on the exterior of the Brixton for things like filters and cables, but the zippered front pocket of the Union Street is better for pens, phones, and more valuable stuff. Personally I like the look of the Union better, with its riveted straps and uninterrupted waxed canvas flap.

If I had to choose, I’d go with the Union Street again, since it’s not so much larger that it becomes cumbrous, but the extra space may make the difference between having to pack a second bag or not.

Filson 24-Hour Tin Briefcase ($395)

Pros: Versatile, well made and guaranteed, spacious

Cons: Lighter material and wax, floppy handles, storm flap nitpick

Filson has been a Seattle standby for a century and more, with its signature waxed-canvas jackets covering the bodies of the hip, the outdoorsy, and the tourists alike. Their most practical bag is this one, the 24-Hour Tin Briefcase, which as the name indicates is a little more on the overnight bag side of things.

This bag has a large main compartment with a padded laptop area that will hold a 15-incher easily, and a couple pockets on the inside to isolate toothbrushes and pens and the like. On the outside is a pair of good-size zippered pockets that open wide to allow access from either the top or side; inside those are organizer strips and sub-pockets for pens and so on.

This is definitely the best generalist out of the bags I tried — it’s equally at home as a daily driver or at the airport. Essentially it’s the perfect “personal item” carry-on. When I’m leaving for a trip I invariably grab this bag because it’s so adaptable. Although it looks a bit bulky it flattens down well when not full, but it doesn’t look weird when it’s packed tightly.

[gallery ids="1660230,1660234,1660231,1660229,1660227,1660226,1660224"]

A bonus with Filson is that should it ever rip or fail — and I mean ever — you can take it in and they’ll fix or patch it for free. I’ve done this with my jackets and it’s 100% awesome. The scars where the tears were make for even more character.

On the other hand, unlike many Filson products this one feels only lightly waxed. If you want more protection from rain you’ll want to add some wax yourself, not something everyone wants to do. You’ll eventually re-wax any of these bags, but this one just seemed to need it right off the bat. The material is a little lighter than some of the other bags, but that could be a plus or a minus. I wouldn’t mind if it was a bit more heavy-duty, like their “rugged twill.”

The handles are nicely made and thick, but tend to sort of flop around when not needed. And the storm flap that covers the top zipper, while welcome, feels like it has the snap on the wrong side — it makes attaching or detaching it a two-hand affair. When it isn’t full, the bag can be a bit shapeless — it’s not really boardroom ready. For that you want Croots or Ernest Alexander below.

Ernest Alexander Walker and Hudson – $385

Pros: Great texture and color, nice style details, low-profile

Cons: Impractical closure on Hudson, Walker has limited space, looks compromise utility a bit

Note: I tried two bags from this maker and unfortunately in the meantime both have sold out. I’ve asked when they’ll be back on the market, but for now you can take this review as a general indicator of the quality of EN bags.

The one I took to from the start is the Walker; it has a pleasantly sleek, minimal look on the outside, the material a handsome chocolate color that has started to wear well. But open up the flap and you have this lovely blue fine canvas inside (there’s a reverse scheme as well). To me this was the most refined of all the bags in this roundup. I like that there are no snaps, clips, or anything visible on the outside — just a wide expanse of that beautiful material.

It’s slim bag but not restrictively so; if what you need to carry isn’t awkward or bulky, there’s room for a good amount in there. Books, a mirrorless with a pancake lens, laptop — sure. But you’re definitely not fitting a spare set of clothes or some groceries.

The small zippered exterior pocket is great for a phone or cables, while the deep interior and exterior pockets are easily accessed and relatively spacious. If you control your loadout, there’s room for lots of stuff in here.

[gallery ids="1660219,1660221,1660218,1660222,1660217,1660216"]

Unfortunately, if you don’t control it, the bag gets bent out of shape easily. Because the top flap attaches to the bottom at the center, if it gets too full the whole thing bulges awkwardly and the tips flip out. And the carry strap, alas, tends to tug on the flap in a way that draws its sides up and away from the clip. And don’t even try to pick it up with the flap detached.

Placing the clip underneath the flap also makes for a fiddly procedure — you have to lift up one side to get at it, and because the loop flips down when not in use, it becomes a two-handed operation to put the two pieces together. A sturdier, more fixed loop would make this easier. But it’s all in the name of style, and the sleek exterior may make up for these fussy aspects.

The cross-body strap has a lot of extra material but I made it into a neat little knot. I think it works pretty well, actually.

The larger Hudson messenger I was prepared to like but ultimately just can’t recommend. Theoretically it’s fantastic, with magnetic pocket closures, tons of room, and a cross between the simplicity of the Walker and the versatility of the Filson bag. But the closure system is just too much of a hassle.

It’s two straps in a simple belt style, which are a huge pain to do over and over if you’re frequently opening and closing the bag. Compared to Ona closures, which combine speed with the flexibility of belt-style adjustment, it just takes forever to access the Hudson. If they make a revised version of this bag that addresses this, it will have my hearty recommendation.

Croots England Vintage Canvas Laptop – $500

Pros: Handsome, well padded, excellent craftsmanship and materials

Cons: Flappy handles, uneven wear, laptop compartment, expensive

Having encountered a Croots bag in the wild one time, I knew I had to include this long-time waxed canvas player in the roundup. Croots waxed canvas is less oily than Filson or ONA, more like a heavy sailcloth. It feels very strong and holds its shape well. It is however on the high end of the spectrum.

That said, because of its stiffness, the Vintage Canvas Laptop bag seems to want to wear prematurely in areas that stick out a bit, like corners or folds near stitching. The wear process shifts the material from the smooth, almost ballistic nylon texture to a rough fuzzy one that I’m not so sure about. The aging from just a couple weeks of use already has me a little worried but it’s also very thick canvas.

The design is a bit more busy than the Ernest Alexander bags, but very handsome and mostly practical. I love the olive color, which contrasts beautifully with the red backing for the zippers. It doesn’t look Christmas-y at all, don’t worry.

The straps are a standout feature. The thick leather handles are attached below the zipper and rear pocket to D-rings, which in turn attach to separate leather straps that go under the entire bag. First this means that the handles flip down easily out of the way, since the D-rings rotate in their loops. The riveted construction also means that there’s no stitching to worry about in the whole strap assembly. And the bottoms of the loops do a little basic protection of the canvas down there.

[gallery ids="1660207,1660206,1660210,1660211,1660212,1660209,1660208,1660204,1660214"]

It also means that when you’re walking, the outside handle tends to flap rather ungracefully against the side; the inner one, up or down, will be rubbing against your flank or back. You can however stow them in the side pockets with a bit of effort, which is a thoughtful touch.

The interior is a lovely shade of red, with several large loose pockets and some stiff leather ones for notebooks and so on. Unfortunately the laptop pocket is poorly proportioned: it’s hugely spacious, enough for three or four laptops to slide in, but the button to snap it shut is so low that I can’t get it fastened over a single 13-inch MacBook Pro. The idea that it could hold a 15-inch is ludicrous.

There’s lots of padding, though, so I wasn’t worried about anything banging around. There’s also the option for a separate camera insert, though large SLR users will likely want to size up.

There isn’t a heck of a lot of room in there but this is definitely meant to be a daily driver briefcase and not an overnight bag — a “personal item” on the plane perhaps but I would take the Filson or ONA over it for space reasons. However as a bag to take to work, the cafe, or the bookstore it’s a great option and a striking one. The Flight Bag is a slightly more expansive and unique option.

S-Zone – $30

Pros: Price, magnetic closures, leather edge details

Cons: Cheap-feeling interior and leather, little padding for laptop

To balance out the admittedly very expensive bags in this review I decided to grab a cheap one off Amazon as well. As I expected, it isn’t up to the quality level of the others, but for $30 it’s a bargain. If you want to experience how waxed canvas evolves and wears, an inexpensive bag like this is a great way to try it out.

The S-Zone’s fabric is a little thin but solid, rather stiff to begin with, but that’s fine — it’ll loosen up as you use the bag. The interior is a cheap-feeling synthetic, however — it’ll work, but you won’t feel like royalty using it.

There’s leather detailing all over, and in some places it feels solid, like the attachments for the shoulder strap and at the corners, where there are big patches that will scuff up nicely. But the handle feels like trouble waiting to happen.

[gallery ids="1660249,1660246,1660244,1660247,1660243,1660248,1660245"]

Instead of a D-ring to allow it to flip down, the leather itself has been loosened up so that it’s extra bendy just above where it attaches. When it’s down, the thin rope around which the handle leather is wrapped is exposed; I can just see this getting soaked, bent, soaked again, bent, and getting weaker and weaker.

The front pockets are a little tight, but I like the little magnetic snaps — they make it easy to open and close them without looking. Just be careful not to stuff too much in there or the snaps won’t hold against the pressure. There’s a good deal of room inside, more than the Croots or Ernest Alexander, but less than the ONA or Filson.

But then there’s the curious design choice to put padding in the divider defining the laptop section, rather than on the outside. And the leather corner pieces stop just short of it! That means the only thing between the corner of your laptop and the ground is the nylon and canvas — and they don’t make for much of a cushion. Though the other bags don’t all have dedicated padding in this area, they do all seem to mitigate it better, and the S-Zone bag puts your laptop in the most danger of hitting the ground.


Hopefully the high prices of most of these items won’t turn

Football matches land on your table thanks to augmented reality

It’s World Cup season, so that means that even articles about machine learning have to have a football angle. Today’s concession to the beautiful game is a system that takes 2D videos of matches and recreates them in 3D so you can watch them on your coffee table (assuming you have some kind of augmented reality setup, which you almost certainly don’t). It’s not as good as being there, but it might be better than watching it on TV.

The “Soccer On Your Tabletop” system takes as its input a video of a match and watches it carefully, tracking each player and their movements individually. The images of the players are then mapped onto 3D models “extracted from soccer video games,” and placed on a 3D representation of the field. Basically they cross FIFA 18 with real life and produce a sort of miniature hybrid.

Considering the source data — two-dimensional, low-resolution, and in motion — it’s a pretty serious accomplishment to reliably reconstruct a realistic and reasonably accurate 3D pose for each player.

Now, it’s far from perfect. One might even say it’s a bit useless. The characters’ positions are estimated, so they jump around a bit, and the ball doesn’t really appear much, so everyone appears to just be dancing around on a field. (That’s on the to-do list.)

But the idea is great, and this is a working if highly limited first shot at it. Assuming the system could ingest a whole game based on multiple angles (it could source the footage directly from the networks), you could have a 3D replay available just minutes after the actual match concluded.

Not only that, but wouldn’t it be cool to be able to gather round a central location and watch the game from multiple angles on it? I’ve always thought one of the worst things about watching sports on TVs is everyone is sitting there staring in one direction, seeing the exact same thing. Letting people spread out, pick sides, see things from different angles to analyze strategies — that would be fantastic.

All we need is for someone to invent a perfect, affordable holographic display that works from all angles and we’re set.

The research is being presented at the Computer Vision and Pattern Recognition conference in Salt Lake City, and it’s a collaboration between Facebook, Google, and the University of Washington.

What’s under those clothes? This system tracks body shapes in real time

With augmented reality coming in hot and depth tracking cameras due to arrive on flagship phones, the time is right to improve how computers track the motions of people they see — even if that means virtually stripping them of their clothes. A new computer vision system that does just that may sound a little creepy, but it definitely has its uses.

The basic problem is that if you’re going to capture a human being in motion, say for a movie or for an augmented reality game, there’s a frustrating vagueness to them caused by clothes. Why do you think motion capture actors have to wear those skintight suits? Because their JNCO jeans make it hard for the system to tell exactly where their legs are. Leave them in the trailer.

Same for anyone wearing a dress, a backpack, a jacket — pretty much anything other than the bare minimum will interfere with the computer getting a good idea of how your body is positioned.

The multi-institutional project (PDF), due to be presented at CVPR in Salt Lake City, combines depth data with smart assumptions about how a body is shaped and what it can do. The result is a sort of X-ray vision, revealing the shape and position of a person’s body underneath their clothes, that works in real time even during quick movements like dancing.

The paper builds on two previous methods, DynamicFusion and BodyFusion. The first uses single-camera depth data to estimate a body’s pose, but doesn’t work well with quick movements or occlusion; the second uses a skeleton to estimate pose but similarly loses track during fast motion. The researchers combined the two approaches into “DoubleFusion,” essentially creating a plausible skeleton from the depth data and then sort of shrink-wrapping it with skin at an appropriate distance from the core.

As you can see above, depth data from the camera is combined with some basic reference imagery of the person to produce both a skeleton and track the joints and terminations of the body. On the right there, you see the results of just DynamicFusion (b), just BodyFusion (c), and the combined method (d).

The results are much better than either method alone, seemingly producing excellent body models from a variety of poses and outfits:

Hoodies, headphones, baggy clothes, nothing gets in the way of the all-seeing eye of DoubleFusion.

One shortcoming, however, is that it tends to overestimate a person’s body size if they’re wearing a lot of clothes — there’s no easy way for it to tell whether someone is broad or they are just wearing a chunky sweater. And it doesn’t work well when the person interacts with a separate object, like a table or game controller — it would likely try to interpret those as weird extensions of limbs. Handling these exceptions is planned for future work.

The paper’s first author is Tao Yu of Tsinghua University in China, but researchers from Beihang University, Google, USC, and the Max Planck Institute were also involved.

“We believe the robustness and accuracy of our approach will enable many applications, especially in AR/VR, gaming, entertainment and even virtual try-on as we also reconstruct the underlying body shape,” write the authors in the paper’s conclusion. “For the first time, with DoubleFusion, users can easily digitize themselves.”

There’s no use denying that there are lots of interesting applications of this technology. But there’s also no use denying that this technology is basically X-ray Spex.

SpeakSee makes it simple for a deaf person to join a group conversation

There’s a great deal of activity in the fields of speech recognition and the “internet of things,” but one natural application of the two has gone relatively unpursued: helping the deaf and hard of hearing take part in everyday conversations. SpeakSee aims to do this (after crowdfunding, naturally) with a clever hardware design that minimizes setup friction and lets everyone communicate naturally.

It’s meant to be used in situations where someone hard of hearing needs to talk with a handful of others — a meeting, a chat at dinner, asking directions and so on. There are speech-to-text apps out there that can transcribe what someone is saying, but they’re not really suited to the purpose.

“Many deaf people experienced a huge barrier in asking people to download the app and hold the phone close to their mouth. These limitations in the interface meant no one kept using it,” explained SpeakSee CEO and co-founder Jari Hazelebach. “But because we designed our own hardware, we were able to customize it towards the situations it will be used in.”

SpeakSee is simple to use: a set of clip-on microphones live in a little charger case, and when the user wants to have a conversation, they hand those microphones out to whoever will be talking. The case acts as a wireless hub for the mics and relays the audio to the smartphone it’s paired with. This audio is sent off, transcribed quickly somewhere in the cloud, and displayed on the deaf user’s phone.

Critically, though, each microphone also intelligently and locally accounts for its speaker and background noise.

“Naturally the microphones pick up speech from multiple people,” said Hazelebach. “So we included sensors that tell the microphone what direction the sound is coming from, and the microphones exchange these values. So we can determine which microphone should pick up which person’s speech.”

The result is quickly transcribed speech divided by speaker, delivered quickly and with decent accuracy (there’s always a trade-off between turnaround time and how the process is). And no one has to do anything but wear a mic. (They have a patent pending for this multi-microphone system.)

Hazelebach’s parents are deaf, and he grew up seeing how their ability to interact in ordinary circumstances was being limited.

The mics aren’t exactly small… but that’s how you know they’re real working hardware and not imaginary.

“As you can imagine my parents were the first to test this out,” he said. “At first we had a lot of issues but soon we started engaging with others. We wrote a post on a deaf blog and out of nowhere 200 people signed up. So we’ve been testing in the field with groups in the U.S., and also in the U.K. and the Netherlands.”

Right now English speech recognition is considerably ahead of Dutch and other languages, so the transcriptions will be better for the former, but even so the devices should work with any of 120 languages supported by the cloud service. Transcription is free for up to 5 hours of audio monthly, after which it’s a $10/month subscription. But if it works, it may be more than worth the money.

The team has a finished prototype but is seeking crowdfunding to get production off the ground. “We need to improve the electronics to meet specifications, battery life for example. We expect to ship in February of 2019,” Hazelebach said. Pre-orders are set at $350 for a dock and three mics.

The usual caveats (primarily “emptor”) apply when backing an Indiegogo type campaign — but at the very least, having spoken to the creator, I feel pretty sure this is a real, working product that just needs a boost to get to market.

Facebook’s new AI research is a real eye-opener

There are plenty of ways to manipulate photos to make you look better, remove red eye or lens flare, and so on. But so far the blink has proven a tenacious opponent of good snapshots. That may change with research from Facebook that replaces closed eyes with open ones in a remarkably convincing manner.

It’s far from the only example of intelligent “in-painting,” as the technique is called when a program fills in a space with what it thinks belongs there. Adobe in particular has made good use of it with its “context-aware fill,” allowing users to seamlessly replace undesired features, for example a protruding branch or a cloud, with a pretty good guess at what would be there if it weren’t.

But some features are beyond the tools’ capacity to replace, one of which is eyes. Their detailed and highly variable nature make it particularly difficult for a system to change or create them realistically.

Facebook, which probably has more pictures of people blinking than any other entity in history, decided to take a crack at this problem.

It does so with a Generative Adversarial Network, essentially a machine learning system that tries to fool itself into thinking its creations are real. In a GAN, one part of the system learns to recognize, say, faces, and another part of the system repeatedly creates images that, based on feedback from the recognition part, gradually grow in realism.

From left to right: “Exemplar” images, source images, Photoshop’s eye-opening algorithm, and Facebook’s method.

In this case the network is trained to both recognize and replicate convincing open eyes. This could be done already, but as you can see in the examples at right, existing methods left something to be desired. They seem to paste in the eyes of the people without much consideration for consistency with the rest of the image.

Machines are naive that way: they have no intuitive understanding that opening one’s eyes does not also change the color of the skin around them. (For that matter, they have no intuitive understanding of eyes, color, or anything at all.)

What Facebook’s researchers did was to include “exemplar” data showing the target person with their eyes open, from which the GAN learns not just what eyes should go on the person, but how the eyes of this particular person are shaped, colored, and so on.

The results are quite realistic: there’s no color mismatch or obvious stitching because the recognition part of the network knows that that’s not how the person looks.

In testing, people mistook the fake eyes-opened photos for real ones, or said they couldn’t be sure which was which, more than half the time. And unless I knew a photo was definitely tampered with, I probably wouldn’t notice if I was scrolling past it in my newsfeed. Gandhi looks a little weird, though.

It still fails in some situations, creating weird artifacts if a person’s eye is partially covered by a lock of hair, or sometimes failing to recreate the color correctly. But those are fixable problems.

You can imagine the usefulness of an automatic eye-opening utility on Facebook that checks a person’s other photos and uses them as reference to replace a blink in the latest one. It would be a little creepy, but that’s pretty standard for Facebook, and at least it might save a group photo or two.

Elizabeth Holmes reportedly steps down at Theranos after criminal indictment

Elizabeth Holmes has left her role as CEO of Theranos and has been charged with wire fraud, CNBC and others report. The company’s former president, Ramesh “Sunny” Balwani, was also indicted today by a grand jury.

These criminal charges are separate from the civil ones filed in March by the SEC and already settled. There are 11 charges; two are conspiracy to commit wire fraud (against investors, and against doctors and patients) and the remaining nine are actual wire fraud, with amounts ranging from the cost of a lab test to $100 million.

Theranos’s general counsel, David Taylor, has been appointed CEO. What duty the position actually entails in the crumbling enterprise is unclear. Holmes, meanwhile, remains chairman of the board.

The FBI Special Agent in Charge of the case against Theranos, John Bennett, said the company engaged in “a corporate conspiracy to defraud financial investors,” and “misled doctors and patients about the reliability of medical tests that endangered health and lives.”

This story is developing. I’ve asked Theranos for comment and will update if I hear back; indeed I’m not even sure anyone is there to respond.

Judge says ‘literal but nonsensical’ Google translation isn’t consent for police search

Machine translation of foreign languages is undoubtedly a very useful thing, but if you’re going for anything more than directions or recommendations for lunch, its shallowness is a real barrier. And when it comes to the law and constitutional rights, a “good enough” translation doesn’t cut it, a judge has ruled.

The ruling (PDF) is not hugely consequential, but it is indicative of the evolving place in which translation apps find themselves in our lives and legal system. We are fortunate to live in a multilingual society, but for the present and foreseeable future it seems humans are still needed to bridge language gaps.

The case in question involved a Mexican man named Omar Cruz-Zamora, who was pulled over by cops in Kansas. When they searched his car, with his consent, they found quite a stash of meth and cocaine, which naturally led to his arrest.

But there’s a catch: Cruz-Zamora doesn’t speak English well, so the consent to search the car was obtained via an exchange facilitated by Google Translate — an exchange that the court found was insufficiently accurate to constitute consent given “freely and intelligently.”

The fourth amendment prohibits unreasonable search and seizure, and lacking a warrant or probable cause, the officers required Cruz-Zamora to understand that he could refuse to let them search the car. That understanding is not evident from the exchange, during which both sides repeatedly fail to comprehend what the other is saying.

Not only that, but the actual translations provided by the app weren’t good enough to accurately communicate the question. For example, the officer asked “¿Puedo buscar el auto?” — the literal meaning of which is closer to “can I find the car,” not “can I search the car.” There’s no evidence that Cruz-Zamora made the connection between this “literal but nonsensical” translation and the real question of whether he consented to a search, let alone whether he understood that he had a choice at all.

With consent invalidated, the search of the car is rendered unconstitutional, and the charges against Cruz-Zamora are suppressed.

It doesn’t mean that consent is impossible via Google Translate or any other app — for example, if Cruz-Zamora had himself opened his trunk or doors to allow the search, that likely would have constituted consent. But it’s clear that app-based interactions are not a sure thing. This will be a case to consider not just for cops on the beat looking to help or investigate people who don’t speak English, but in courts as well.

Providers of machine translation services would have us all believe that those translations are accurate enough to use in most cases, and that in a few years they will replace human translators in all but the most demanding situations. This case suggests that machine translation can fail even the most basic tests, and as long as that possibility remains, we have to maintain a healthy skepticism.

Machines learn language better by using a deep understanding of words

Computer systems are getting quite good at understanding what people say, but they also have some major weak spots. Among them is the fact that they have trouble with words that have multiple or complex meanings. A new system called ELMo adds this critical context to words, producing better understanding across the board.

To illustrate the problem, think of the word “queen.” When you and I are talking and I say that word, you know from context whether I’m talking about Queen Elizabeth, or the chess piece, or the matriarch of a hive, or RuPaul’s Drag Race.

This ability of words to have multiple meanings is called polysemy. And really, it’s the rule rather than the exception. Which meaning it is can usually be reliably determined by the phrasing — “God save the queen!” versus “I saved my queen!” — and of course all this informs the topic, the structure of the sentence, whether you’re expected to respond, and so on.

Machine learning systems, however, don’t really have that level of flexibility. The way they tend to represent words is much simpler: it looks at all those different definitions of the word and comes up with a sort of average — a complex representation, to be sure, but not reflective of its true complexity. When it’s critical that the correct meaning of a word gets through, they can’t be relied on.

A new method called ELMo (“Embeddings from Language Models”), however, lets the system handle polysemy with ease; as evidence of its utility, it was awarded best paper honors at NAACL last week. At its heart it uses its training data (a huge collection of text) to determine whether a word has multiple meanings and how those different meanings are signaled in language.

For instance, you could probably tell in my example “queen” sentences above, despite their being very similar, that one was about royalty and the other about a game. That’s because the way they are written contain clues to your own context-detection engine to tell you which queen is which.

Informing a system of these differences can be done by manually annotating the text corpus from which it learns — but who wants to go through millions of words making a note on which queen is which?

“We were looking for a method that would significantly reduce the need for human annotation,” explained Mathew Peters, lead author of the paper. “The goal was to learn as much as we can from unlabeled data.”

In addition, he said, traditional language learning systems “compress all that meaning for a single word into a single vector. So we started by questioning the basic assumption: let’s not learn a single vector, let’s have an infinite number of vectors. Because the meaning is highly dependent on the context.”

ELMo learns this information by ingesting the full sentence in which the word appears; it would learn that when a king is mentioned alongside a queen, it’s likely royalty or a game, but never a beehive. When it sees pawn, it knows that it’s chess; jack implies cards; and so on.

An ELMo-equipped language engine won’t be nearly as good as a human with years of experience parsing language, but even working knowledge of polysemy is hugely helpful in understanding a language.

Not only that, but taking the whole sentence into account in the meaning of a word also allows the structure of that sentence to be mapped more easily, automatically labeling clauses and parts of speech.

Systems using the ELMo method had immediate benefits, improving on even the latest natural language algorithms by as much as 25 percent — a huge gain for this field. And because it is a better, more context-aware style of learning, but not a fundamentally different one, it can be integrated easily even into existing commercial systems.

In fact, Microsoft is reportedly already using it with Bing. After all, it’s crucial in search to determine intention, which of course requires an accurate reading of the query. ELMo is open source, too, like all the work from the Allen Institute for AI, so any company with natural language processing needs should probably check this out.

The paper lays down the groundwork of using ELMo for English language systems, but because its power is derived by essentially a close reading of the data that it’s fed, there’s no theoretical reason why it shouldn’t be applicable not just for other languages, but in other domains. In other words, if you feed it a bunch of neuroscience texts, it should be able to tell the difference between temporal as it relates to time and as it relates to that region of the brain.

This is just one example of how machine learning and language are rapidly developing around each other; although it’s already quite good enough for basic translation, speech to text and so on, there’s quite a lot more that computers could do via natural language interfaces — if they only know how.