My school record of 2 hours to write the ceiling of 20 dishes identify java program.

1.2. Baidu Cloud Smart Profile

1.2.1 Baidu image recognition service

Baidu image recognition services, based on deep learning and large-scale image training, comprehensive information to accurately identify the object in the picture category, location, and so on confidence.

1.2.2 Recognition dishes

Support identify 50,000 kinds of dishes adapted to recognize only dishes containing a single picture, the interface returns the name of the dishes, confidence, calories, Wikipedia information and other comprehensive information. Support self dishes Gallery for identifying images containing multiple dishes, the interface returns the name of the dishes, location, relevance and other comprehensive information.

1.2.3 recognition application scenarios dishes

According to take a picture, the picture identification names of dishes, the dishes get a reference caloric content and Wikipedia information can be combined with the recognition results provide further dietary recommendations, health-related functions such as program management, enhance the user experience, widely used in food and beverage and entertainment Health Management APP in. According to take a picture, identify the name and location of the picture dishes, improve settlement efficiency and reduce manual entry cost, widely used in intelligent settlement in the catering industry.

1.2.4 Introduction recognition interface dishes

Interface capability: the request for identification dishes. I.e., for an image (normally decoded, and an appropriate aspect ratio) of the input, name of the output image of the dishes, calorie information, confidence.

Request format: POST mode call (Content-Type is application / x-www-form-urlencoded, and then through the request body urlencode formatting)

Returned Format: JSON format

Request limit: base64 encoded subject to request image: base64 encoded picture refers to a picture encoded data string into a string, the string is used instead of the image address. First, you can get a binary image, and then use Base64 encoding format can be. (Base64 encoded picture is not included in the picture header, such as (data: image / jpg; base64), a request format support: PNG, JPG, JPEG, BMP, GIF **, the size does not exceed base64 encoded 4M, the smallest side at least 15px, the longest side maximum 4096px, an aspect ratio of 3: 1 or less)

1.2.5 Request Interface Parameter Description

parameter	Are Required	Types of	The values range	Explanation
image	Yes	string	-	Base64 encoded string, mandatory request to the image file format. (Supported image formats: jpg, bmp, png, jpeg ), image size no more than 4M. The shortest side at least 15px, the longest side maximum 4096px. Note: The image needs to base64 encoded, remove the head and then perform encoding urlencode .
top_num	no	unit32	-	Return result top n, default 5.
filter_threshold	Yes	float	-	0.95 default can be adjusted to identify the effect of this parameter, the recognition rate decreased non-food.
baike_num	no	integer	0	The number of results returned encyclopedic information, default does not return

1.2.6 Interface Parameter Description Returns

Field	Are Required	Types of	Explanation
log_id	Yes	uint64	The only log id, used to locate the problem
result_num	no	unit32	The number of results to return the number of elements, and in the result array
result	no	array()	Dishes recognition result array
+name	no	string	Dishes, example: fish-flavored pork
+calorie	no	float	Calories, calorie content per 100g
+probability	no	float	Recognition result confidence value for each row, 0-1
+ Baike_info	object	no	Corresponding to the recognition result Wikipedia entry name
++ baike_url	string	no	Corresponding to the recognition result of Baidu encyclopedia pages link
++image_url	string	no	The corresponding recognition results Encyclopedia image link
++description	string	no	Corresponding to the recognition result Encyclopedia Description