mark/image-similarity
1
0
Fork 0
Code Pull Requests Packages Projects Releases Activity

wasm module. linter updates
continuous-integration/drone/push Build is failing
Details

Browse Source
This commit is contained in:
Mark Hoekveen
2024-06-25 17:21:28 +02:00
parent 5e6a6c62bf
commit fd616f37fd
12 changed files with 108 additions and 461 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/store.rs
+30 -71
View File
@@ -34,8 +34,7 @@ impl PRStats {
};
let tpos = self.true_positives[t] as f64;
let fpos = self.false_positives[t] as f64;
let p = tpos / (tpos + fpos);
p
tpos / (tpos + fpos)
}
pub fn recall(&self, threshold: usize) -> f64 {
let t = match threshold {
@@ -44,8 +43,7 @@ impl PRStats {
};
let tpos = self.true_positives[t] as f64;
let fneg = self.false_negatives[t] as f64;
let r = tpos / (tpos + fneg);
r
tpos / (tpos + fneg)
}
pub fn pr(&self, threshold: usize) -> (f64, f64) {
let t = match threshold {
@@ -87,7 +85,7 @@ pub struct DescriptorStore {
}
impl DescriptorStore {
/// Makes a new empty DescriptorStore with default settings
/// Makes a new empty `DescriptorStore` with default settings
pub fn new(descriptor: Box<dyn Descriptor>) -> Self {
let map: HashMap<u64, Vec<String>> = HashMap::new();
//let seen: HashSet<String> = HashSet::new();
@@ -105,23 +103,22 @@ impl DescriptorStore {
pub fn with_file<P: AsRef<Path>>(mut self, path: P) -> Self {
self.save_location = std::path::PathBuf::from(path.as_ref());
let map_file = fs::read(&self.save_location);
match map_file {
Ok(f) => {
self.map = rmp_serde::from_slice(&f).unwrap();
for i in self.map.keys() {
self.bktree.insert(*i);
}
let mut seen: HashSet<String> = HashSet::new();
for bucket in self.map.values() {
for element in bucket {
if !element.starts_with("mut.") {
seen.insert(element.clone());
}
if let Ok(f) = map_file {
self.map = rmp_serde::from_slice(&f).unwrap();
for i in self.map.keys() {
self.bktree.insert(*i);
}
let mut seen: HashSet<String> = HashSet::new();
for bucket in self.map.values() {
for element in bucket {
if !element.starts_with("mut.") {
seen.insert(element.clone());
}
}
self.seen = Some(seen);
},
Err(_) => info!("{} not found, starting from empty store.", self.save_location.display()),
}
self.seen = Some(seen);
} else {
info!("{} not found, starting from empty store.", self.save_location.display());
};
self
}
@@ -147,8 +144,8 @@ impl DescriptorStore {
Ok(value) => value,
Err(_e) => return Err(SaveError::Serialization),
};
match fs::write(&self.save_location, &serialized) {
Ok(_) => Ok(()),
match fs::write(&self.save_location, serialized) {
Ok(()) => Ok(()),
Err(_e) => Err(SaveError::File),
}
}
@@ -167,7 +164,7 @@ impl DescriptorStore {
},
None => {
for bucket in self.map.values() {
if bucket.contains(&value) {
if bucket.contains(value) {
return true;
}
}
@@ -178,7 +175,7 @@ impl DescriptorStore {
pub fn get(&self, value: &String) -> Option<u64> {
for (key, bucket) in self.map.iter() {
if bucket.contains(&value) {
if bucket.contains(value) {
return Some(*key);
}
}
@@ -187,7 +184,7 @@ impl DescriptorStore {
/// Inserts a single value into the store
pub fn insert(&mut self, key: &u64, value: String) {
let bucket = match self.map.get(&key) {
let bucket = match self.map.get(key) {
Some(b) => {
let mut n = b.clone();
if !n.contains(&value) {
@@ -213,7 +210,9 @@ impl DescriptorStore {
continue
}
};
if !self.has_value(&name) { // !self.contains(name.to_string()) {
if !self.has_value(&name) {
debug!("{} already known, skipping.", name);
} else {
info!("Processing {}", name);
let img = match image::open(file.path()) {
Ok(v) => v,
@@ -227,8 +226,6 @@ impl DescriptorStore {
println!("{} duplicate of {:?}", name, self.map.get(&phash));
}
self.insert(&phash, name);
} else {
debug!("{} already known, skipping.", name);
}
self.save().expect("error");
}
@@ -238,15 +235,15 @@ impl DescriptorStore {
pub fn store(&mut self, img: &DynamicImage, name: String) {
if !self.has_value(&name) {
let phash = self.descriptor.describe(&img);
let phash = self.descriptor.describe(img);
self.insert(&phash, name);
}
}
/// Nearest neighbours
pub fn nn(&self, from: u64, max_distance: usize) -> Vec<(&u64, isize)> {
let mut neighbours = self.bktree.find(from, max_distance.try_into().unwrap());
neighbours.sort_by(|a, b| a.1.cmp(&b.1));
let neighbours = self.bktree.find(from, max_distance.try_into().unwrap());
//neighbours.sort_by(|a, b| a.1.cmp(&b.1));
neighbours
}
@@ -263,44 +260,6 @@ impl DescriptorStore {
results
}
pub fn print_nn(&self, from: u64, max_distance: usize, show_images: bool) {
//println!("{from:064b}");
let neighbours = self.nn(from, max_distance);
let (term_width, _) = viuer::terminal_size();
let mut x = 0;
let mut y = 8;
for (element, distance) in neighbours {
let elements = self.map.get(element);
match elements {
Some(paths) => {
for path in paths {
println!("{element:064b}: {:?} (distance: {distance})", path);
if show_images {
if x+16 >= term_width {
x = 0;
y += 8;
}
let conf = viuer::Config {
width: Some(16),
height: Some(8),
x,
y,
use_kitty: false,
..Default::default()
};
x += 16;
let path = "data/".to_string() + path;
let img = image::open(&path).unwrap();
let img = img.grayscale().thumbnail_exact(8, 8);
viuer::print(&img, &conf).expect("Image printing failed.");
}
}
},
None => ()
};
}
}
pub fn print_most_dups(&self) {
let mut most = 0;
for bucket in self.map.values() {
@@ -319,7 +278,7 @@ impl DescriptorStore {
let value_entry = format!("{value}\n");
output.push_str(&value_entry);
}
output.push_str("\n");
output.push('\n');
}
}
print!("DescriptorStore:\n{}Total: {}", output, self.map.len())
@@ -415,7 +374,7 @@ impl fmt::Display for DescriptorStore {
let value_entry = format!("\t{value}\n");
output.push_str(&value_entry);
}
output.push_str("\n");
output.push('\n');
}
write!(f, "DescriptorStore:\n{}Total: {}", output, self.map.len())
}