31 results about "Loss weight" patented technology

The invention provides a face attribute recognition method of a deep neural network based on cascaded multi-task learning and relates to the computer vision technology. Firstly, a cascaded deep convolutional neural network is designed, then multi-task learning is used for each cascaded sub-network in the cascaded deep convolutional neural network, four tasks of face classification, border regression, face key point detection and face attribute analysis are learned simultaneously, then a dynamic loss weighting mechanism is used in the deep convolutional neural network based on the cascaded multi-task learning to calculate the loss weights of face attributes, finally a face attribute recognition result of a last cascaded sub-network is used as the final face attribute recognition result based on a trained network model. A cascading method is used to jointly train three different sub-networks, end-to-end training is achieved, the result of face attribute recognition is optimized, different from the use of fixed loss weights in a loss function, a difference between the face attributes present is considered in the invention.

An identification method of the rhizome gastrodiae includes the below processes: (1) setting the standard chemical fingerprint atlas of rhizome gastrodiae: a. putting the rhizome gastrodiae standard sample into the sorbitic extractor; b. adding the ethanol and putting in the water bath for hot reflux, treating it by the ultrasonic cell knapper and weight, then to loss weight by the corresponding solvent; repeating it again; after filtration, concentration and drying, it uses the 10% acetonitrile to constant volume; c. using the HPLC, Eclipse, XD13-C8 chromatographic column, the 10% acetonitrile solution is as the mobile phase, the light of 221+-1nm wavelength is as the detection light source to determine the chemical fingerprint atlas of rhizome gastrodiae of rhizome gastrodiae standard sample; which has seven evidence peaks, the peak with 1.920min holding time is the rhizome gastrodiae element peak, the rhizome gastrodiae element peak of the other rhizome gastrodiae has the big warp of 1.9% with the standard sample. (2) Determining the chemical fingerprint of the detected sample and the dummy sample according to the progress (1) and comparing to identify.

The invention discloses a test method for corrosion-resistance performance contrast of a lead-acid battery grid, comprising the steps of (1) taking several kinds of grids, each for two pieces, soaked in the sugar alkali solution to remove surface oxide, weighing after cleaning & drying, records the weight of the monolithic grid under test, remembers as w1; (2) placing the grid under test respectively in the corresponding corrosion device containing sulfuric acid solution; (3) connecting all the corrosion device in series, and charged with constant current; (4) after the charge, takes out the grid under test which represents the anode, weighs after processing according to the method of step (1) , remembers as w2; (5) adopts the formula w = w1-W2 to calculate the corrosion loss weight for each grid under test, and compares the corrosion resistance performance of several kind of grids under test based on the quantities of w. The invention takes the grid weight of removing surface oxide layer as the initial weight, makes the last calculated corrosion loss weight more accurately to reflect the corrosion resistance performance of the grid under test.

The invention provides a landslide debris flow area detection method based on sparse representation classification. The method includes: dividing a target area image to be detected into multiple image blocks to be detected with the same sizes, analyzing and judging the brightness of three channels of red, green and blue through a sparse representation classification method, obtaining a suspected landslide debris flow image block collection, presetting the mean value of the difference between the first threshold value and the sparse representation errors; obtaining the landslide debris flow severely damaged area and the area boundaries in the suspected landslide debris flow image block collection through a connected area detection method; obtaining the land type of the image of each point before a disaster according to the position of each point of the area boundaries, obtaining the expansion degree influence coefficient of each point corresponds to the disaster loss weight according to the preset land type, obtaining the extended distance of the landslide debris flow area according to the mean value of the difference between the preset first threshold value and the sparse representation errors, and obtaining the landslide debris flow general damaged area through the extension of the landslide debris flow seriously damaged area. The landslide debris flow detection method is high in detection efficiency and accuracy and low in false alarm rate.

A method for an attention-driven image segmentation by using at least one adaptive loss weight map is provided to be used for updating HD maps required to satisfy level 4 of autonomous vehicles. By this method, vague objects such as lanes and road markers at distance may be detected more accurately. Also, this method can be usefully performed in military, where identification of friend or foe is important, by distinguishing aircraft marks or military uniforms at distance. The method includes steps of: a learning device instructing a softmax layer to generate softmax scores; instructing a loss weight layer to generate loss weight values by applying loss weight operations to predicted error values generated therefrom; and instructing a softmax loss layer to generate adjusted softmax loss values by referring to initial softmax loss values, generated by referring to the softmax scores and their corresponding GTs, and the loss weight values.

The invention discloses a glutinous rice cake chilli seasoning and a production process thereof. The formula and contents of glutinous rice cake chilli seasoning include 100 kg of dried chili, 8-12 kg of garlic, 7-10 kg of ginger, an appropriate amount of salt, an appropriate amount of high quality Daqu liquor, an appropriate amount of tsaoko. The production processes of the glutinous rice cake chilli include: cleaning the dried chili after removing stems and impurities, immersing in warm water for 3-5 hours, adding the peeled garlic, the peeled ginger, the Daqu liquor, the salt and the tsaoko, triturating to 10-20 meshes, sterilizing, testing, bagging and sealing. The product is fragrant in smell, has the nature of acridity and warmth, is spicy but not fierce, is rich in flavor, invigorates spleen and warms middle, is anti-inflammatory and diuretic, and helps to improve the body resistance, while can sterilize, fend off cold, lower lipid, help to loss weight and supply vitamin C.

The invention discloses an intestine-opening and lung-moistening ferment and a preparation method thereof. The intestine-opening and lung-moistening ferment comprises the following raw materials in parts by weight: 10-15 parts of pumpkins, 15-20 parts of unpolished rice, 6-8 parts of white fungi, 5-10 parts of edible amaranth, 5-8 parts of chenopodium album, 8-10 parts of watermelon rind, 5-8 parts of tangerine peel, 3-5 parts of walnut kernels, 5-8 parts of bananas and 5-8 parts of honey. By adopting scientific matches of the raw materials, the prepared ferment is capable of decomposing and eliminating the toxins in the intestinal tract and in the body so as to achieve the effects of beautifying and keeping youth; the ferment is also capable of readjusting the balance performance of the human body so as to promote massive decomposition of the deposited fat to achieve the effect of reducing weight; moreover, the ferment further has the functions of eliminating dryness and moistening the lung. Thus, the intestine-opening and lung-moistening ferment is especially suitable for people with obesity or constipation. Furthermore, the intestine-opening and lung-moistening ferment is also rich in nutrient contents, and capable of satisfying demands of the body. The consumers of the ferment can loss weight without malnutrition.

The invention discloses a vehicle weighing state detection system, which comprises weight counting equipment, a weight counting charging system, a signal shunt, a front-end collecting module and a rear-end early warning module. The signal shunt collects analog signals from the weight counting equipment, and the analog signals are imaged in the signal shunt by an emitter follower to form two signals equal to the original signals in magnitude, wherein one way of the signals is returned to the weight counting charging system, and the other way of signal is returned to the front-end collecting module to be subjected to data analysis; the front-end collecting module conducts data analysis on each signal, and returns the calculation results to the rear-end early warning module; and the rear-endearly warning module can directly calculate an equipment failure module, sends out a reminding repairing signal, and automatically compensates the loss total weight to the system. An owner monitoringplatform is effectively combined, the failure position can be quickly determined, meanwhile, the weight counting result after data repair of damaged equipment is given, and owners are assisted in counting the loss weight.

The present invention discloses a quality control method of asthma tablets, which includes the flowing steps: (1) chromatography conditions and system suitability experiment; (2) preparation of control solution; (3) preparation of test solution is as following: taking the product, removing the coating, porphyrizing, taking about 1g, placing into a 50ml volumetric flask, adding 1ml of ammonia test solution and 25ml of chloroform precisely, weighing the weight, carrying out the ultrasonic treatment with the power of 260w and the frequency of 40KHz for 30 minutes, cooling, weighing the weight, using the chloroform to supplement the loss weight, shaking evenly, filtering, sucking 5ml of filtrate precisely, adding 2ml of 2 percent hydrochloric ethanol anhydrous, shaking evenly, drying up by evaporation in a water bath, adding 10ml of 0.1percent hydrochloric acid in the residual slag, full dissolving by ultrasound, filtering, and taking the subsequent filtrate; (4) weigh and suck the control solution and the test solution precisely and respectively to inject to an HPLC instrument for determination. The method adopts the HPLC to carry out the content determination of morphine, codeine phosphate and ephedrine hydrochloride in the prescription under a same system, which can not only ensure the efficacy of the products, but can also enable the quality of the products to be stable.

The invention belongs to the technical field of image recognition, and particularly relates to a face recognition multi-path deep neural network method based on adaptive weight, and the technical scheme comprises the following steps: firstly, constructing a multi-path neural network suitable for face recognition; secondly, establishing a corresponding multi-path loss function based on a task of face recognition; finally, using an adaptive weight algorithm for training, wherein the algorithm adaptively adjusts loss weights of different path networks according to corresponding thresholds in thetraining process, and a final model is obtained. According to the invention, the multi-path neural network is established; according to the invention, an end-to-end face recognition process is completed, training of a deep network is carried out through an adaptive weight algorithm, the training effect is significantly improved, the method can also be suitable for faces of different distances, theperformance of practical application is greatly improved, the market demand can be well met, and large-area popularization and application of the method in the market are facilitated.

The invention belongs to the technical field of food and in particular relates to a nutritious breakfast and a preparation method thereof. The nutritious breakfast comprises the following ingredients in parts by weight: 4-8 parts of black soy beans, 2-4 parts of red dates, 2-4 parts of peanuts, 4-8 parts of soya beans, 1-2 parts of walnuts, 1-2 parts of eggs, 1-1.8 parts of kelp, 0.5-1 part of abalone, 0.4-0.8 part of corn starch, 0.5-0.9 part of gelatin, 1-2 parts of sterculia nobilis, 1-2 parts of indochina dragonplum fruit, 0.2-0.7 part of perilla leaf, 0.3-0.6 part of fresh ginger, 0.5-2 parts of lemon, 0.2-0.3 part of lactic acid bacteria and 0.1-0.2 part of beta-cyclodextrin. The nutritious breakfast contains various nutrient elements and is simple and convenient to eat and convenient to carry. The nutritious breakfast has rich and balanced nutrition and meets the physiological needs of a human body in the morning. Meanwhile, the nutritious breakfast also has the effect of reducing fat so as to help women to loss weight. The nutritious breakfast has varied edible methods and is unique in taste and can greatly improve appetite.

The method for measuring alloy property boiling point-weight has method adopted the following steps: using precision balance to weight small quantity of tested alloy powder, placing the alloy powder into a porcelain boat, weighing total weight of alloy powder and porcelain boat, placing them into a porcelain pipe in high-temp. furnace, under the condition of argon protection and constant pressurequickly-heating to predefined temp., heat-insualting for 10-30 min., then quickly-cooling to room temp., then weight the total weight of porcelain boat and residual alloy powder and resolving evaporated quantity, according to the evaporated loss weight quantity measured at several temp. points calcualting evaporated loss weight percentage content and drawing temp.-evaporated percentage content curve.

The invention discloses a traffic illuminating system which comprises a lamp pole and a lamp holder which are fixedly connected. A watering device, a water pump and an infrared induction device used for detecting the human body are arranged on the lamp pole, the end, close to the lamp holder, of the lamp pole is rotationally connected with the watering device with the lamp pole as the axis, the water pump is communicated with the watering device, the infrared induction device is coupled with the water pump, when the human body is detected by the infrared induction device, an induction signal is transmitted to the water pump, watering is stopped after the induction signal is received by the water pump, the watering device is rotationally connected to the lamp pole, and therefore the watering device rotates to achieve large-area afforesting through spraying; and as the watering device is arranged at the end, close to the lamp holder, of the lamp pole, water can be sprayed out from the high position, water drops loss weight when falling from the high position, water can be further scattered, the weight of water can be scattered, and impact to afforesting is reduced.

The invention associates to a method preparing supercritical fiber, which includes island component polymer and sea component polymer spinning supercritical fiber. The key point is that island component is polyolefin polymer and sea component is a polymer which is different in solution. The method offers a way to produce a kind of superfine fiber base cloth. Using superfine fiber produced in that way we can get a sort of backing material, and then let sea component of backing material solute to get the superfine fiber base cloth. Because of the feature that island component made from polyolefin polymer is of high bending elasticity and low density, the the superfine fiber base cloth produced in this way can loss weight very much and be up to the thickness when sea component solute.

The invention provides a dynamic weight measuring method for a steel ball mill, wherein a B-P function characteristic curve among a cylinder body total weight B (comprising milled object weight and steel ball weight), a high-pressure lubricant pressure P and a contact area S of the steel ball mill can be obtained through establishing a basic model and a true model. According to the high-pressure lubricant pressure P which is measured in real time in operation, the real-time cylinder body weight B can be obtained through reverse computing. The dynamic weight measuring method can conveniently and quickly detect the real-time dynamic weight of the steel ball mill, thereby calculating steel loss weight, obtaining a steel loss curve, and correctly guiding personnel to add steel balls. Simultaneously through comparing the steel loss of steel balls and liner plates which are in different brands or are made of different materials, the dynamic weight measuring method can easily determine which material or brand of the liner plates or steel balls is better. Furthermore a real-time ball to powder ratio can be detected, so that the steel ball mill is always kept in an optimal pulverizing state, thereby greatly improving powder the pulverizing efficiency of the steel ball mill, greatly reducing power consumption of the steel ball mill, and truly realizing energy reduction and production increase.

The invention belongs to the field of fruit and vegetable fresh-keeping storage technologies, and discloses a fresh-keeping storage method of garlic. The method comprises the following steps: placing garlic in a sealing bag; sealing the sealing bag, and placing the bag of garlic in a cold storage for refrigerating and preserving. Advantages of the present invention are that: the expiration date of the garlic stored with the method provided by the present invention can be prolonged to about two years, the garlic does not loss weight during the preservation period, original mouthfeel of the garlic can be effectively maintained, and the operation of the storage processes is simple.