Skip to content

Commit a51accc

Browse files
gDreamcatchergDreamcatcher
authored
Update README.md
1 parent 8328652 commit a51accc

1 file changed

Lines changed: 159 additions & 0 deletions

File tree

‎README.md‎

Lines changed: 159 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,162 @@
1+
# Quickstart
2+
3+
*This Quickstart suit for those who want to search something but do not know how to extract image or text to features. Others refer to [APILowLevel.md](APILowLevel.md) .*
4+
5+
Vearch is aimed to build a simple and fast image retrieval system. Through this system, a image retrieval system could be easily built, including image object detection, feature extraction and similarity search. This quickstart demonstrates how to use it.
6+
7+
![images/main_process.gif](images/main_process.gif)
8+
9+
10+
11+
## Before you begin
12+
13+
1. Deploy Vearch system referred to [Deploy.md](https://github.com/vearch/vearch/blob/master/docs/Deploy.md).
14+
15+
For testing you can download [coco data](https://pjreddie.com/media/files/val2014.zip), or use the images in images folder we choose from [coco data](https://pjreddie.com/media/files/val2014.zip). For more details, you can refer test folder in `plugin.src`
16+
17+
## Different from APILowLevel.md
18+
19+
This API is similar to [APILowLevel.md](https://github.com/vearch/vearch/blob/master/docs/APILowLevel.md), and plugin can perfectly adapt to it, Any method defined in APILowLevel.md can be used by plugin. However, if vector has existed, the APILowLevel.md is the fisrt choice.
20+
21+
The difference:
22+
23+
- The name of db can not be one of ['_cluster', 'list', 'db', 'space'].
24+
- Can not use `_msearch` method.
25+
- Replace the feature field with the object requiring the feature, refer to insert or search demo.
26+
27+
28+
## Deploy a plugin service
29+
30+
This requires only two operations:
31+
32+
1. Modify parameters in `src/config.py`;
33+
2. Execution script:
34+
For image, `bash ./bin/run.sh image` ;
35+
For video, `bash ./bin/run.sh video`;
36+
For text, `bash ./bin/run.sh text` ;
37+
38+
39+
## Create a database and space
40+
41+
Before inserting and searching, a database and space should be creating firstly. Use the following `curl` command to create a new database and space.
42+
43+
```shell
44+
# create a db which name test
45+
curl -XPUT -H "content-type:application/json" -d '{"name": "test"}' http://127.0.0.1:4101/db/_create
46+
47+
# create a space in test db which name test too.
48+
## for image
49+
curl -XPUT -H "content-type: application/json" -d '{"name": "test", "partition_num": 1, "replica_num": 1, "engine": {"name": "gamma", "index_size": 70000, "max_size": 10000000, "id_type": "String", "retrieval_type": "IVFPQ", "retrieval_param": { "metric_type": "InnerProduct", "ncentroids": 256, "nsubvector": 32 } }, "properties": { "url": { "type": "keyword", "index": true }, "feature1": { "type": "vector", "dimension":512, "format": "normalization" } }} }' http://127.0.0.1:4101/space/test/_create
50+
51+
## for text dimension is 768
52+
curl -XPUT -H "content-type: application/json" -d '{"name": "test", "partition_num": 1, "replica_num": 1, "engine": {"name": "gamma", "index_size": 70000, "max_size": 10000000, "id_type": "String", "retrieval_type": "IVFPQ", "retrieval_param": { "metric_type": "InnerProduct", "ncentroids": 256, "nsubvector": 32 } }, "properties": { "text": { "type": "keyword", "index": true }, "feature1": { "type": "vector", "dimension":768, "format": "normalization" } }} }' http://127.0.0.1:4101/space/test/_create
53+
```
54+
55+
A successful response looks like this:
56+
57+
```shell
58+
# create db
59+
{"code":200,"msg":"success","data":{"id":1,"name":"test"}}
60+
61+
# create space
62+
{"code":200,"msg":"success","data":{"id":2,"name":"test","version":2,"db_id":1,"enabled":true,"partitions":[{"id":3,"space_id":2,"db_id":1,"partition_slot":0,"replicas":[1]},{"id":4,"space_id":2,"db_id":1,"partition_slot":2147483647,"replicas":[1]}],"partition_num":2,"replica_num":1,"properties":{ "url": { "type": "keyword", "index":true}, "feature1": { "type": "vector", "dimension":512, "format": "normalization" }},"engine":{"name":"gamma","index_size":100000,"max_size":100000,"nprobe":-1,"metric_type":"InnerProduct","ncentroids":-1,"nsubvector":-1,"nbits_per_idx":-1}}}
63+
```
64+
65+
66+
67+
## Delete a database and space
68+
69+
If the database and space is not need again, Use the following `curl` command to delete a database and space.
70+
71+
```shell
72+
73+
curl -XDELETE http://127.0.0.1:4101/space/test/test
74+
curl -XDELETE http://127.0.0.1:4101/db/test
75+
```
76+
77+
A successful response looks like this:
78+
79+
```shell
80+
{"code":200,"msg":"success"}
81+
```
82+
83+
84+
85+
## Insert data into space
86+
87+
Use the following `curl` command to insert single data into space.
88+
89+
The method of single import demo:
90+
91+
```shell
92+
# single insert
93+
94+
## image
95+
curl -XPOST -H "content-type: application/json" -d' { "url": "../images/COCO_val2014_000000123599.jpg", "feature1":{"feature":"../images/COCO_val2014_000000123599.jpg"}} ' http://127.0.0.1:4101/test/test/AW63W9I4JG6WicwQX_RC
96+
97+
## text
98+
curl -XPOST -H "content-type: application/json" -d' { "text": "感谢大家", "feature1":{"feature":"感谢大家"} } ' http://127.0.0.1:4101/test/test/AW63W9I4JG6WicwQX_RC'
99+
```
100+
101+
A successful response like this:
102+
103+
```shell
104+
{"_index":"test","_type":"test","_id":"AW63W9I4JG6WicwQX_RC","status":201,"_version":1,"_shards":{"total":0,"successful":1,"failed":0},"result":"created","_seq_no":1,"_primary_term":1}
105+
```
106+
107+
## Get record by ID
108+
109+
Use the following `curl` command to get a record by ID
110+
111+
```shell
112+
# request
113+
curl -XGET http://127.0.0.1:4101/test/test/AW63W9I4JG6WicwQX_RC
114+
115+
# response
116+
{"_index":"test","_type":"test","_id":"AW63W9I4JG6WicwQX_RC","found":true,"_version":1,"_source":{"url":"../images/COCO_val2014_000000123599.jpg"}}
117+
```
118+
119+
120+
121+
## Delete record by ID
122+
123+
Use the following `curl` command to delete a record by ID
124+
125+
```shell
126+
# request
127+
curl -XDELETE http://127.0.0.1:4101/test/test/AWz2IFBSJG6WicwQVTog
128+
129+
# response
130+
{"_index":"test","_type":"test","_id":"AW63W9I4JG6WicwQX_RC","status":200,"_version":0,"_shards":{"total":0,"successful":1,"failed":0},"result":"unknow","_seq_no":1,"_primary_term":1}
131+
```
132+
133+
134+
135+
136+
## Search similar result from space
137+
138+
Searching by an image URI for an publicly accessible online image or an image stored in images folders.
139+
140+
Searching by an image stored in images folders or image URI on Internet, using the following `curl` command to search similar result from space
141+
142+
```shell
143+
curl -H "content-type: application/json" -XPOST -d '{ "query": { "sum": [{"feature":"../images/COCO_val2014_000000123599.jpg", "field":"feature1"}]}}' http://127.0.0.1:4101/test/test/_search
144+
145+
```
146+
147+
A successful response looks like this:
148+
149+
```shell
150+
{"took":14,"timed_out":false,"_shards":{"total":2,"failed":0,"successful":2},"hits":{"total":1,"max_score":0.9999997615814209,"hits":[{"_index":"test","_type":"test","_id":"AW8OftTLJG6WicwQyAt2","_score":0.9999997615814209,"_extra":{"vector_result":[{"field":"feature1","source":"","score":0.9999997615814209}]},"_version":1,"_source":{"url":"../images/COCO_val2014_000000123599.jpg"}}]}}
151+
```
152+
153+
search result look like this
154+
155+
![images/COCO_val2014_000000123599.jpg](images/COCO_val2014_000000123599.jpg)
156+
157+
![images/result.jpg](images/result.jpg)
158+
159+
1160
# Docker
2161
3162
* build: `docker build . -t vearch/plugin:latest`

0 commit comments

Comments
 (0)